Machine Learning Detects Pan-cancer Ras Pathway Activation in The Cancer Genome Atlas

Comprehensive data mining reveals RTK/RAS signaling pathway as a promoter of prostate cancer lineage plasticity through transcription factors and CNV

Prostate cancer lineage plasticity is a key driver in the transition to neuroendocrine prostate cancer (NEPC), and the RTK/RAS signaling pathway is a well-established cancer pathway. Nevertheless, the comprehensive link between the RTK/RAS signaling pathway and lineage plasticity has received limited investigation. In particular, the intricate regulatory network governing the interplay between RTK/RAS and lineage plasticity remains largely unexplored. The multi-omics data were clustered with the coefficient of argument and neighbor joining algorithm. Subsequently, the clustered results were analyzed utilizing the GSEA, gene sets related to stemness, multi-lineage state datasets, and canonical cancer pathway gene sets. Finally, a comprehensive exploration of the data based on the ssGSEA, WGCNA, GSEA, VIPER, prostate cancer scRNA-seq data, and the GPSAdb database was conducted. Among the six modules in the clustering results, there are 300 overlapping genes, including 3 previously unreported prostate cancer genes that were validated to be upregulated in prostate cancer through RT-qPCR. Function Module 6 shows a positive correlation with prostate cancer cell stemness, multi-lineage states, and the RTK/RAS signaling pathway. Additionally, the 19 leading-edge genes of the RTK/RAS signaling pathway promote prostate cancer lineage plasticity through a complex network of transcriptional regulation and copy number variations. In the transcriptional regulation network, TP63 and FOXO1 act as suppressors of prostate cancer lineage plasticity, whereas RORC exerts a promoting effect. This study provides a comprehensive perspective on the role of the RTK/RAS pathway in prostate cancer lineage plasticity and offers new clues for the treatment of NEPC.

A deep profile of gene expression across 18 human cancers

Clinically and biologically valuable information may reside untapped in large cancer gene expression data sets. Deep unsupervised learning has the potential to extract this information with unprecedented efficacy but has thus far been hampered by a lack of biological interpretability and robustness. Here, we present DeepProfile, a comprehensive framework that addresses current challenges in applying unsupervised deep learning to gene expression profiles. We use DeepProfile to learn low-dimensional latent spaces for 18 human cancers from 50,211 transcriptomes. DeepProfile outperforms existing dimensionality reduction methods with respect to biological interpretability. Using DeepProfile interpretability methods, we show that genes that are universally important in defining the latent spaces across all cancer types control immune cell activation, while cancer type-specific genes and pathways define molecular disease subtypes. By linking DeepProfile latent variables to secondary tumor characteristics, we discover that tumor mutation burden is closely associated with the expression of cell cycle-related genes. DNA mismatch repair and MHC class II antigen presentation pathway expression, on the other hand, are consistently associated with patient survival. We validate these results through Kaplan-Meier analyses and nominate tumor-associated macrophages as an important source of survival-correlated MHC class II transcripts. Our results illustrate the power of unsupervised deep learning for discovery of novel cancer biology from existing gene expression data.

A multiomics analysis-assisted deep learning model identifies a macrophage-oriented module as a potential therapeutic target in colorectal cancer

Colorectal cancer (CRC) is a common malignancy involving multiple cellular components. The CRC tumor microenvironment (TME) has been characterized well at single-cell resolution. However, a spatial interaction map of the CRC TME is still elusive. Here, we integrate multiomics analyses and establish a spatial interaction map to improve the prognosis, prediction, and therapeutic development for CRC. We construct a CRC immune module (CCIM) that comprises FOLR2+ macrophages, exhausted CD8+ T cells, tolerant CD8+ T cells, exhausted CD4+ T cells, and regulatory T cells. Multiplex immunohistochemistry is performed to depict the CCIM. Based on this, we utilize advanced deep learning technology to establish a spatial interaction map and predict chemotherapy response. CCIM-Net is constructed, which demonstrates good predictive performance for chemotherapy response in both the training and testing cohorts. Lastly, targeting FOLR2+ macrophage therapeutics is used to disrupt the immunosuppressive CCIM and enhance the chemotherapy response in vivo.

Quantified pathway mutations associate epithelial-mesenchymal transition and immune escape with poor prognosis and immunotherapy resistance of head and neck squamous cell carcinoma

BACKGROUND

Pathway mutations have been calculated to predict the poor prognosis and immunotherapy resistance in head and neck squamous cell carcinoma (HNSCC). To uncover the unique markers predicting prognosis and immune therapy response, the accurate quantification of pathway mutations are required to evaluate epithelial-mesenchymal transition (EMT) and immune escape. Yet, there is a lack of score to accurately quantify pathway mutations.

MATERIAL AND METHODS

Firstly, we proposed Individualized Weighted Hallmark Gene Set Mutation Burden (IWHMB, https://github.com/YuHongHuang-lab/IWHMB ) which integrated pathway structure information and eliminated the interference of global Tumor Mutation Burden to accurately quantify pathway mutations. Subsequently, to further elucidate the association of IWHMB with EMT and immune escape, support vector machine regression model was used to identify IWHMB-related transcriptomic features (IRG), while Adversarially Regularized Graph Autoencoder (ARVGA) was used to further resolve IRG network features. Finally, Random walk with restart algorithm was used to identify biomarkers for predicting ICI response.

RESULTS

We quantified the HNSCC pathway mutation signatures and identified pathway mutation subtypes using IWHMB. The IWHMB-related transcriptomic features (IRG) identified by support vector machine regression were divided into 5 communities by ARVGA, among which the Community 1 enriching malignant mesenchymal components promoted EMT dynamically and regulated immune patterns associated with ICI responses. Bridge Hub Gene (BHG) identified by random walk with restart was key to IWHMB in EMT and immune escape, thus, more predictive for ICI response than other 70 public signatures.

CONCLUSION

In summary, the novel pathway mutation scoring-IWHMB suggested that the elevated malignancy mediated by pathway mutations is a major cause of poor prognosis and immunotherapy failure in HNSCC, and is capable of identifying novel biomarkers to predict immunotherapy response.

Optimizer's dilemma: optimization strongly influences model selection in transcriptomic prediction

Motivation

Most models can be fit to data using various optimization approaches. While model choice is frequently reported in machine-learning-based research, optimizers are not often noted. We applied two different implementations of LASSO logistic regression implemented in Python's scikit-learn package, using two different optimization approaches (coordinate descent, implemented in the liblinear library, and stochastic gradient descent, or SGD), to predict mutation status and gene essentiality from gene expression across a variety of pan-cancer driver genes. For varying levels of regularization, we compared performance and model sparsity between optimizers.

Results

After model selection and tuning, we found that liblinear and SGD tended to perform comparably. liblinear models required more extensive tuning of regularization strength, performing best for high model sparsities (more nonzero coefficients), but did not require selection of a learning rate parameter. SGD models required tuning of the learning rate to perform well, but generally performed more robustly across different model sparsities as regularization strength decreased. Given these tradeoffs, we believe that the choice of optimizers should be clearly reported as a part of the model selection and validation process, to allow readers and reviewers to better understand the context in which results have been generated.

Availability and implementation

The code used to carry out the analyses in this study is available at https://github.com/greenelab/pancancer-evaluation/tree/master/01_stratified_classification. Performance/regularization strength curves for all genes in the Vogelstein et al. (2013) dataset are available at https://doi.org/10.6084/m9.figshare.22728644.

A pan-cancer analysis of the prognostic implication and oncogenic role of tubulin epsilon and delta complex 2 (TEDC2) in human tumors

Introduction

Tubulin epsilon and delta complex 2 (TEDC2) is widely expressed in various human tissues and primarily governs centriole stability. However, the biological significance of TEDC2 in pan-cancer is unclear.

Methods

In this study, we employed R software and various online bioinformatics analysis tools to investigate the functional attributes of TEDC2 in human tumours and its potential involvement in immune response. The status of TEDC2 expression was evaluated in samples from the TCGA and GEO datasets, as well as in tumour and corresponding normal samples from the TCGA database. Subsequently, Kaplan-Meier estimates, clinical correlations, and univariate Cox regressions were used to analyze the 33 types of tumors from TCGA and determine the prognostic significance of TEDC2. Moreover, nomogram models were formulated using three distinct tumours, namely kidney renal clear cell carcinoma (KIRC), lung adenocarcinoma (LUAD), and liver hepatocellular carcinoma (LIHC), to evaluate the prognostic significance of TEDC2 in tumours. Furthermore, TEDC2 was investigated for its correlation with the levels of immune cell infiltration, and a functional enrichment analysis was conducted to identify potential signalling pathways involving TEDC2.

Results

Differential analysis revealed that 16 tumour types expressed TEDC2 to a greater extent than normal tissues. The abnormal expression of TEDC2 can predict survival outcomes in patients with adrenocortical carcinoma (ACC), KIRC, kidney renal papillary cell carcinoma (KIRP), LUAD, LIHC, lower grade glioma (LGG), and thymoma (THYM). Subsequent results indicated that TEDC2 has the ability to influence ECM regulators, cell cycle, and Immune checkpoint-associated signalling pathways, which could potentially lead to a poor prognosis and tumour progression.

Discussion

TEDC2 has been identified as a potential therapeutic target that could predict the prognosis of multiple tumour types, making it a promising target for reversing tumour development.

Deep learning-enabled breast cancer endocrine response determination from H&E staining based on ESR1 signaling activity

Estrogen receptor (ER) positivity by immunohistochemistry has long been a main selection criterium for breast cancer patients to be treated with endocrine therapy. However, ER positivity might not directly correlate with activated ER signaling activity, which is a better predictor for endocrine therapy responsiveness. In this study, we investigated if a deep learning method using whole-slide H&E-stained images could predict ER signaling activity. First, ER signaling activity score was determined using RNAseq data available from each of the 1082 breast cancer samples in the TCGA Pan-Cancer dataset based on the Hallmark Estrogen Response Early gene set from the Molecular Signature Database (MSigDB). Then the processed H&E-stained images and ER signaling activity scores from a training cohort were fed into ResNet101 with three additional fully connected layers to generate a predicted ER activity score. The trained models were subsequently applied to an independent testing cohort. The result demonstrated that ER + /HER2- breast cancer patients with a higher predicted ER activity score had longer progression-free survival (p = 0.0368) than those with lower predicted ER activity score. In conclusion, a convolutional deep neural network can predict prognosis and endocrine therapy response in breast cancer patients based on whole-slide H&E-stained images. The trained models were found to robustly predict the prognosis of ER + /HER2- patients. This information is valuable for patient management, as it does not require RNA-seq or microarray data analyses. Thus, these models can reduce the cost of the diagnosis workflow if such information is required.

Identifying Explainable Machine Learning Models and a Novel SFRP2+ Fibroblast Signature as Predictors for Precision Medicine in Ovarian Cancer

Ovarian cancer (OC) is a type of malignant tumor with a consistently high mortality rate. The diagnosis of early-stage OC and identification of functional subsets in the tumor microenvironment are essential to the development of patient management strategies. However, the development of robust models remains unsatisfactory. We aimed to utilize artificial intelligence and single-cell analysis to address this issue. Two independent datasets were screened from the Gene Expression Omnibus (GEO) database and processed to obtain overlapping differentially expressed genes (DEGs) in stage II-IV vs. stage I diseases. Three explainable machine learning algorithms were integrated to construct models that could determine the tumor stage and extract important characteristic genes as diagnostic biomarkers. Correlations between cancer-associated fibroblast (CAF) infiltration and characteristic gene expression were analyzed using TIMER2.0 and their relationship with survival rates was comprehensively explored via the Kaplan-Meier plotter (KM-plotter) online database. The specific expression of characteristic genes in fibroblast subsets was investigated through single-cell analysis. A novel fibroblast subset signature was explored to predict immune checkpoint inhibitor (ICI) response and oncogene mutation through Tumor Immune Dysfunction and Exclusion (TIDE) and artificial neural network algorithms, respectively. We found that Support Vector Machine-Shapley Additive Explanations (SVM-SHAP), Extreme Gradient Boosting (XGBoost), and Random Forest (RF) successfully diagnosed early-stage OC (stage I). The area under the receiver operating characteristic curves (AUCs) of these models exceeded 0.990. Their overlapping characteristic gene, secreted frizzled-related protein 2 (SFRP2), was a risk factor that affected the overall survival of OC patients with stage II-IV disease (log-rank test: p < 0.01) and was specifically expressed in a fibroblast subset. Finally, the SFRP2+ fibroblast signature served as a novel predictor in evaluating ICI response and exploring pan-cancer tumor protein P53 (TP53) mutation (AUC = 0.853, 95% confidence interval [CI]: 0.829-0.877). In conclusion, the models based on SVM-SHAP, XGBoost, and RF enabled the early detection of OC for clinical decision making, and SFRP2+ fibroblast signature used in diagnostic models can inform OC treatment selection and offer pan-cancer TP53 mutation detection.

Targeting oncogenic TERT promoter variants by allele-specific epigenome editing

BACKGROUND

Activation of dominant oncogenes by small or structural genomic alterations is a common driver mechanism in many cancers. Silencing of such dominantly activated oncogenic alleles, thus, is a promising strategy to treat cancer. Recently, allele-specific epigenome editing (ASEE) has been described as a means to reduce transcription of genes in an allele-specific manner. In cancer, specificity to an oncogenic allele can be reached by either targeting directly a pathogenic single-nucleotide variant or a polymorphic single-nucleotide variant linked to the oncogenic allele. To investigate the potential of ASEE in cancer, we here explored this approach by targeting variants at the TERT promoter region. The TERT promoter region has been described as one of the most frequently mutated non-coding cancer drivers.

RESULTS

Sequencing of the TERT promoter in cancer cell lines showed 53% (41/77) to contain at least one heterozygous sequence variant allowing allele distinction. We chose the hepatoblastoma cell line Hep-G2 and the lung cancer cell line A-549 for this proof-of-principle study, as they contained two different kinds of variants, namely the activating mutation C228T in the TERT core promoter and the common SNP rs2853669 in the THOR region, respectively. These variants were targeted in an allele-specific manner using sgRNA-guided dCas9-DNMT3A-3L complexes. In both cell lines, we successfully introduced DNA methylation specifically to the on-target allele of the TERT promoter with limited background methylation on the off-target allele or an off-target locus (VEGFA), respectively. We observed a maximum CpG methylation gain of 39% and 76% on the target allele when targeting the activating mutation and the common SNP, respectively. The epigenome editing translated into reduced TERT RNA expression in Hep-G2.

CONCLUSIONS

We applied an ASEE-mediated approach to silence TERT allele specifically. Our results show that the concept of dominant oncogene inactivation by allele-specific epigenome editing can be successfully translated into cancer models. This new strategy may have important advantages in comparison with existing therapeutic approaches, e.g., targeting telomerase, especially with regard to reducing adverse side effects.

Unraveling the relationship between the renin-angiotensin system and endometrial cancer: a comprehensive review

Endometrial cancer (EC), the most common adenocarcinoma, represents 90% of uterine cancer in women with an increased incidence of occurrence attributed to age, obesity, hypertension, and hypoestrogenism. Being the most common gynecological malignancy in women, it shows a relation with the activation of different components of the renin-angiotensin system (RAS), which is predominantly involved in maintaining blood pressure, salt, water, and aldosterone secretion, thereby playing a significant role in the etiology of hypertension. The components of the RAS, i.e., ACE-I, ACE-II, AT1R, AT2R, and Pro(renin) receptor, are widely expressed in both glandular and stromal cells of the endometrium, with varying levels throughout the different phases of the menstrual cycle. This causes the endometrial RAS to implicate angiogenesis, neovascularization, and cell proliferation. Thus, dysfunctioning of the endometrial RAS could predispose the growth and spread of EC. Interestingly, the increased expression of AngII, AGTR1, and AGTR2 showed advancement in the stages and progression of EC via the prorenin/ATP6AP2 and AngII/AGTR1 pathway. Therefore, this review corresponds to unraveling the relationship between the progression and development of endometrial cancer with the dysfunction in the expression of various components associated with RAS in maintaining blood pressure.

Identification of cancer types from gene expressions using learning techniques

Tumor is the major cause of death all around the world in recent days. Early detection and prediction of a cancer type are important for a patient's well-being. Functional genomic data has recently been used in the effective and early detection of cancer. According to previous research, the use of microarray data in cancer prediction has evidenced two main problems as high dimensionality and limited sample size. Several researchers have used numerous statistical and machine learning-based methods to classify cancer types but still, limitations are there which makes cancer classification a difficult job. Deep Learning (DL) and Convolutional Neural Networks (CNN) have been proven with effective analyses of unstructured data including gene expression data. In the proposed method gene expression data for five types of cancer is collected from The Cancer Genome Atlas (TCGA). Prominent features are selected using a hybrid Particle Swarm Optimization (PSO) and Random Forest (RF) algorithm followed by the use of Principal Component Analysis (PCA) for dimensionality reduction. Finally, for classification blend of Convolutional Neural Network (CNN) and Bi-directional Long Short Term Memory (Bi-LSTM) is used to predict the target type of cancer. Experimental results demonstrate that accuracy of the proposed method is 96.89%. As compared to existing work, our method outperformed with better results.

A systematic pan-cancer analysis identifies TRIM28 as an immunological and prognostic predictor and involved in immunotherapy resistance

Tripartite motif-containing protein 28 (TRIM28), as a transcriptional cofactor, has pleiotropic biological effects, such as silencing genes, promoting cellular proliferation and differentiation, and facilitating DNA repair. It is reported that TRIM28 is also correlated with immune infiltration in liver cancer that highlights an unnoticed function of TRIM28 in immune system. However, the prognostic and immunotherapeutic role of TRIM28 in human cancer has not been elucidated. In this study, we conducted a systematic pan-cancer analysis and partial experimental validation of TRIM28 as an immunological and prognostic predictor and its involvement in immunotherapy resistance. We found that TRIM28 expression was higher in various tumor tissues than in normal tissues. Higher TRIM28 expression was associated with poorer prognosis in multiple cancers. The expression of TRIM28 was positively correlated with the presence of T cells, macrophages and neutrophils, and TRIM28 also promoted the infiltration of a series of immune cell. Moreover, TRIM28 affected a wide range of cancer-related scores, and the abnormal expression of TRIM28 was also involved in tumor mutational burden, drug sensitivity, and microsatellite instability in cancer. The results suggest that TRIM28 is a potentially valuable immune response indicator and a molecular biomarker for predicting the prognosis of cancer patients.

Unsupervised meta-clustering identifies risk clusters in acute myeloid leukemia based on clinical and genetic profiles

BACKGROUND

Increasingly large and complex biomedical data sets challenge conventional hypothesis-driven analytical approaches, however, data-driven unsupervised learning can detect inherent patterns in such data sets.

METHODS

While unsupervised analysis in the medical literature commonly only utilizes a single clustering algorithm for a given data set, we developed a large-scale model with 605 different combinations of target dimensionalities as well as transformation and clustering algorithms and subsequent meta-clustering of individual results. With this model, we investigated a large cohort of 1383 patients from 59 centers in Germany with newly diagnosed acute myeloid leukemia for whom 212 clinical, laboratory, cytogenetic and molecular genetic parameters were available.

RESULTS

Unsupervised learning identifies four distinct patient clusters, and statistical analysis shows significant differences in rate of complete remissions, event-free, relapse-free and overall survival between the four clusters. In comparison to the standard-of-care hypothesis-driven European Leukemia Net (ELN2017) risk stratification model, we find all three ELN2017 risk categories being represented in all four clusters in varying proportions indicating unappreciated complexity of AML biology in current established risk stratification models. Further, by using assigned clusters as labels we subsequently train a supervised model to validate cluster assignments on a large external multicenter cohort of 664 intensively treated AML patients.

CONCLUSIONS

Dynamic data-driven models are likely more suitable for risk stratification in the context of increasingly complex medical data than rigid hypothesis-driven models to allow for a more personalized treatment allocation and gain novel insights into disease biology.

An integrated model for predicting KRAS dependency

The clinical approvals of KRAS G12C inhibitors have been a revolutionary advance in precision oncology, but response rates are often modest. To improve patient selection, we developed an integrated model to predict KRAS dependency. By integrating molecular profiles of a large panel of cell lines from the DEMETER2 dataset, we built a binary classifier to predict a tumor's KRAS dependency. Monte Carlo cross validation via ElasticNet within the training set was used to compare model performance and to tune parameters α and λ. The final model was then applied to the validation set. We validated the model with genetic depletion assays and an external dataset of lung cancer cells treated with a G12C inhibitor. We then applied the model to several Cancer Genome Atlas (TCGA) datasets. The final "K20" model contains 20 features, including expression of 19 genes and KRAS mutation status. In the validation cohort, K20 had an AUC of 0.94 and accurately predicted KRAS dependency in both mutant and KRAS wild-type cell lines following genetic depletion. It was also highly predictive across an external dataset of lung cancer lines treated with KRAS G12C inhibition. When applied to TCGA datasets, specific subpopulations such as the invasive subtype in colorectal cancer and copy number high pancreatic adenocarcinoma were predicted to have higher KRAS dependency. The K20 model has simple yet robust predictive capabilities that may provide a useful tool to select patients with KRAS mutant tumors that are most likely to respond to direct KRAS inhibitors.

Prevalence, causes and impact of TP53-loss phenocopying events in human tumors

BACKGROUND

TP53 is a master tumor suppressor gene, mutated in approximately half of all human cancers. Given the many regulatory roles of the corresponding p53 protein, it is possible to infer loss of p53 activity - which may occur due to alterations in trans - from gene expression patterns. Several such alterations that phenocopy p53 loss are known, however additional ones may exist, but their identity and prevalence among human tumors are not well characterized.

RESULTS

We perform a large-scale statistical analysis on transcriptomes of ~ 7,000 tumors and ~ 1,000 cell lines, estimating that 12% and 8% of tumors and cancer cell lines, respectively, phenocopy TP53 loss: they are likely deficient in the activity of the p53 pathway, while not bearing obvious TP53 inactivating mutations. While some of these cases are explained by amplifications in the known phenocopying genes MDM2, MDM4 and PPM1D, many are not. An association analysis of cancer genomic scores jointly with CRISPR/RNAi genetic screening data identified an additional common TP53-loss phenocopying gene, USP28. Deletions in USP28 are associated with a TP53 functional impairment in 2.9-7.6% of breast, bladder, lung, liver and stomach tumors, and have comparable effect size to MDM4 amplifications. Additionally, in the known copy number alteration (CNA) segment harboring MDM2, we identify an additional co-amplified gene (CNOT2) that may cooperatively boost the TP53 functional inactivation effect of MDM2. An analysis of cancer cell line drug screens using phenocopy scores suggests that TP53 (in)activity commonly modulates associations between anticancer drug effects and various genetic markers, such as PIK3CA and PTEN mutations, and should thus be considered as a drug activity modifying factor in precision medicine. As a resource, we provide the drug-genetic marker associations that differ depending on TP53 functional status.

CONCLUSIONS

Human tumors that do not bear obvious TP53 genetic alterations but that phenocopy p53 activity loss are common, and the USP28 gene deletions are one likely cause.

Landscape of Genetic Alterations Underlying Hallmark Signature Changes in Cancer Reveals TP53 Aneuploidy-driven Metabolic Reprogramming

The hallmark signatures based on gene expression capture core cancer processes. Through a pan-cancer analysis, we describe the overview of hallmark signatures across tumor types/subtypes and reveal significant relationships between these signatures and genetic alterations. TP53 mutation exerts diverse changes, including increased proliferation and glycolysis, which are closely mimicked by widespread copy-number alterations. Hallmark signature and copy-number clustering identify a cluster of squamous tumors and basal-like breast and bladder cancers with elevated proliferation signatures, frequent TP53 mutation, and high aneuploidy. In these basal-like/squamous TP53-mutated tumors, a specific and consistent spectrum of copy-number alterations is preferentially selected prior to whole-genome duplication. Within Trp53-null breast cancer mouse models, these copy-number alterations spontaneously occur and recapitulate the hallmark signature changes observed in the human condition. Together, our analysis reveals intertumor and intratumor heterogeneity of the hallmark signatures, uncovering an oncogenic program induced by TP53 mutation and select aneuploidy events to drive a worsened prognosis.

Significance

Our data demonstrate that TP53 mutation and a resultant selected pattern of aneuploidies cause an aggressive transcriptional program including upregulation of glycolysis signature with prognostic implications. Importantly, basal-like breast cancer demonstrates genetic and/or phenotypic changes closely related to squamous tumors including 5q deletion that reveal alterations that could offer therapeutic options across tumor types regardless of tissue of origin.

A Bayesian model for unsupervised detection of RNA splicing based subtypes in cancers

Identification of cancer sub-types is a pivotal step for developing personalized treatment. Specifically, sub-typing based on changes in RNA splicing has been motivated by several recent studies. We thus develop CHESSBOARD, an unsupervised algorithm tailored for RNA splicing data that captures "tiles" in the data, defined by a subset of unique splicing changes in a subset of patients. CHESSBOARD allows for a flexible number of tiles, accounts for uncertainty of splicing quantification, and is able to model missing values as additional signals. We first apply CHESSBOARD to synthetic data to assess its domain specific modeling advantages, followed by analysis of several leukemia datasets. We show detected tiles are reproducible in independent studies, investigate their possible regulatory drivers and probe their relation to known AML mutations. Finally, we demonstrate the potential clinical utility of CHESSBOARD by supplementing mutation based diagnostic assays with discovered splicing profiles to improve drug response correlation.

Genomic Landscapes and Hallmarks of Mutant RAS in Human Cancers

The RAS family of small GTPases represents the most commonly activated oncogenes in human cancers. To better understand the prevalence of somatic RAS mutations and the compendium of genes that are coaltered in RAS-mutant tumors, we analyzed targeted next-generation sequencing data of 607,863 mutations from 66,372 tumors in 51 cancer types in the AACR Project GENIE Registry. Bayesian hierarchical models were implemented to estimate the cancer-specific prevalence of RAS and non-RAS somatic mutations, to evaluate co-occurrence and mutual exclusivity, and to model the effects of tumor mutation burden and mutational signatures on comutation patterns. These analyses revealed differential RAS prevalence and comutations with non-RAS genes in a cancer lineage-dependent and context-dependent manner, with differences across age, sex, and ethnic groups. Allele-specific RAS co-mutational patterns included an enrichment in NTRK3 and chromatin-regulating gene mutations in KRAS G12C-mutant non-small cell lung cancer. Integrated multiomic analyses of 10,217 tumors from The Cancer Genome Atlas (TCGA) revealed distinct genotype-driven gene expression programs pointing to differential recruitment of cancer hallmarks as well as phenotypic differences and immune surveillance states in the tumor microenvironment of RAS-mutant tumors. The distinct genomic tracks discovered in RAS-mutant tumors reflected differential clinical outcomes in TCGA cohort and in an independent cohort of patients with KRAS G12C-mutant non-small cell lung cancer that received immunotherapy-containing regimens. The RAS genetic architecture points to cancer lineage-specific therapeutic vulnerabilities that can be leveraged for rationally combining RAS-mutant allele-directed therapies with targeted therapies and immunotherapy.

SIGNIFICANCE

The complex genomic landscape of RAS-mutant tumors is reflective of selection processes in a cancer lineage-specific and context-dependent manner, highlighting differential therapeutic vulnerabilities that can be clinically translated.

Identification of phenocopies improves prediction of targeted therapy response over DNA mutations alone

DNA mutations in specific genes can confer preferential benefit from drugs targeting those genes. However, other molecular perturbations can “phenocopy” pathogenic mutations, but would not be identified using standard clinical sequencing, leading to missed opportunities for other patients to benefit from targeted treatments. We hypothesized that RNA phenocopy signatures of key cancer driver gene mutations could improve our ability to predict response to targeted therapies, despite not being directly trained on drug response. To test this, we built gene expression signatures in tissue samples for specific mutations and found that phenocopy signatures broadly increased accuracy of drug response predictions in-vitro compared to DNA mutation alone, and identified additional cancer cell lines that respond well with a positive/negative predictive value on par or better than DNA mutations. We further validated our results across four clinical cohorts. Our results suggest that routine RNA sequencing of tumors to identify phenocopies in addition to standard targeted DNA sequencing would improve our ability to accurately select patients for targeted therapies in the clinic.

wenda_gpu: fast domain adaptation for genomic data

MOTIVATION

Domain adaptation allows for the development of predictive models even in cases with limited sample data. Weighted elastic net domain adaptation specifically leverages features of genomic data to maximize transferability but the method is too computationally demanding to apply to many genome-sized datasets.

RESULTS

We developed wenda_gpu, which uses GPyTorch to train models on genomic data within hours on a single GPU-enabled machine. We show that wenda_gpu returns comparable results to the original wenda implementation, and that it can be used for improved prediction of cancer mutation status on small sample sizes than regular elastic net.

AVAILABILITY AND IMPLEMENTATION

wenda_gpu is available on GitHub at https://github.com/greenelab/wenda_gpu/.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

RAS oncogenic activity predicts response to chemotherapy and outcome in lung adenocarcinoma

Activating mutations in KRAS occur in 32% of lung adenocarcinomas (LUAD). Despite leading to aggressive disease and resistance to therapy in preclinical studies, the KRAS mutation does not predict patient outcome or response to treatment, presumably due to additional events modulating RAS pathways. To obtain a broader measure of RAS pathway activation, we developed RAS84, a transcriptional signature optimised to capture RAS oncogenic activity in LUAD. We report evidence of RAS pathway oncogenic activation in 84% of LUAD, including 65% KRAS wild-type tumours, falling into four groups characterised by coincident alteration of STK11/LKB1, TP53 or CDKN2A, suggesting that the classifications developed when considering only KRAS mutant tumours have significance in a broader cohort of patients. Critically, high RAS activity patient groups show adverse clinical outcome and reduced response to chemotherapy. Patient stratification using oncogenic RAS transcriptional activity instead of genetic alterations could ultimately assist in clinical decision-making.

Widespread redundancy in -omics profiles of cancer mutation states

BACKGROUND

In studies of cellular function in cancer, researchers are increasingly able to choose from many -omics assays as functional readouts. Choosing the correct readout for a given study can be difficult, and which layer of cellular function is most suitable to capture the relevant signal remains unclear.

RESULTS

We consider prediction of cancer mutation status (presence or absence) from functional -omics data as a representative problem that presents an opportunity to quantify and compare the ability of different -omics readouts to capture signals of dysregulation in cancer. From the TCGA Pan-Cancer Atlas that contains genetic alteration data, we focus on RNA sequencing, DNA methylation arrays, reverse phase protein arrays (RPPA), microRNA, and somatic mutational signatures as -omics readouts. Across a collection of genes recurrently mutated in cancer, RNA sequencing tends to be the most effective predictor of mutation state. We find that one or more other data types for many of the genes are approximately equally effective predictors. Performance is more variable between mutations than that between data types for the same mutation, and there is little difference between the top data types. We also find that combining data types into a single multi-omics model provides little or no improvement in predictive ability over the best individual data type.

CONCLUSIONS

Based on our results, for the design of studies focused on the functional outcomes of cancer mutations, there are often multiple -omics types that can serve as effective readouts, although gene expression seems to be a reasonable default option.

Exploring Histological Similarities Across Cancers From a Deep Learning Perspective

Histopathology image analysis is widely accepted as a gold standard for cancer diagnosis. The Cancer Genome Atlas (TCGA) contains large repositories of histopathology whole slide images spanning several organs and subtypes. However, not much work has gone into analyzing all the organs and subtypes and their similarities. Our work attempts to bridge this gap by training deep learning models to classify cancer vs. normal patches for 11 subtypes spanning seven organs (9,792 tissue slides) to achieve high classification performance. We used these models to investigate their performances in the test set of other organs (cross-organ inference). We found that every model had a good cross-organ inference accuracy when tested on breast, colorectal, and liver cancers. Further, high accuracy is observed between models trained on the cancer subtypes originating from the same organ (kidney and lung). We also validated these performances by showing the separability of cancer and normal samples in a high-dimensional feature space. We further hypothesized that the high cross-organ inferences are due to shared tumor morphologies among organs. We validated the hypothesis by showing the overlap in the Gradient-weighted Class Activation Mapping (GradCAM) visualizations and similarities in the distributions of nuclei features present within the high-attention regions.

Subclass-specific Prognosis and Treatment Efficacy Inference in Head and Neck Squamous Carcinoma

Exploring the prognostic classification and biomarkers in Head and Neck Squamous Carcinoma (HNSC) is of great clinical significance. We hybridized three prominent strategies to comprehensively characterize the molecular features of HNSC. We constructed a 15-gene signature to predict patients death risk with an average AUC of 0.744 for 1-, 3-, and 5-year on TCGA-HNSC training set, and average AUCs of 0.636, 0.584, 0.755 in GSE65858, GSE-112026, CPTAC-HNSCC datasets, respectively. By combined with NMF clustering and consensus clustering of fraction of tumor immune cell infiltration (ICI) in the tumor microenvironment (TME), we captured a more refined biological characteristics of HNSC, and observed a prognosis heterogeneity in high tumor immunity patients. By matching tumor subset-specific expression signatures to drug-induced cell line expression profiles from large-scale pharmacogenomic databases in the OCTAD workspace, we identified a group of HNSC patients featured with poor prognosis and demonstrated that the individuals in this group are likely to receive increased drug sensitivity to reverse differentially expressed disease signature genes. This trend is especially highlighted among those with higher death risk and tumour immunity.

EPIMUTESTR: a nearest neighbor machine learning approach to predict cancer driver genes from the evolutionary action of coding variants

Discovering rare cancer driver genes is difficult because their mutational frequency is too low for statistical detection by computational methods. EPIMUTESTR is an integrative nearest-neighbor machine learning algorithm that identifies such marginal genes by modeling the fitness of their mutations with the phylogenetic Evolutionary Action (EA) score. Over cohorts of sequenced patients from The Cancer Genome Atlas representing 33 tumor types, EPIMUTESTR detected 214 previously inferred cancer driver genes and 137 new candidates never identified computationally before of which seven genes are supported in the COSMIC Cancer Gene Census. EPIMUTESTR achieved better robustness and specificity than existing methods in a number of benchmark methods and datasets.

Genomic Landscape and Clinical Utility in Korean Advanced Pan-Cancer Patients from Prospective Clinical Sequencing: K-MASTER Program

The fundamental principle of precision oncology is centralized on the identification of therapeutically exploitable targets that provides individual patients with cancer an opportunity to make informed decisions on a personalized level. To facilitate and adopt such concepts within clinical practice, we have initiated a nationwide, multi-institutional precision oncology screening program to examine and enroll patients into the most appropriate clinical trial based on their tumor's unique molecular properties. To determine the prevalence of essential major driver mutations and to explore their dynamic associations at both molecular and pathway levels, we present a comprehensive overview on the genomic properties of East Asian patients with cancer. We further delineate the extent of genomic diversity as well as clinical actionability in patients from Western and Eastern cultures at the pan-cancer and single-tumor entity levels. To support fellow oncology communities in future investigations involving large-scale analysis, all data have been made accessible to the public (https://kmportal.or.kr).

SIGNIFICANCE

We present a comprehensive overview of molecular properties of East Asian pan-cancer patients and demonstrate significant diversity in terms of genomic characteristics as well as clinical utility compared with patients with European ancestry. The results of this study will lay the groundwork for designing personalized treatments in the clinical setting. See related commentary by Moyers and Subbiah, p. 886. This article is highlighted in the In This Issue feature, p. 873.

Machine learning in medical applications: A review of state-of-the-art methods

Applications of machine learning (ML) methods have been used extensively to solve various complex challenges in recent years in various application areas, such as medical, financial, environmental, marketing, security, and industrial applications. ML methods are characterized by their ability to examine many data and discover exciting relationships, provide interpretation, and identify patterns. ML can help enhance the reliability, performance, predictability, and accuracy of diagnostic systems for many diseases. This survey provides a comprehensive review of the use of ML in the medical field highlighting standard technologies and how they affect medical diagnosis. Five major medical applications are deeply discussed, focusing on adapting the ML models to solve the problems in cancer, medical chemistry, brain, medical imaging, and wearable sensors. Finally, this survey provides valuable references and guidance for researchers, practitioners, and decision-makers framing future research and development directions.

Dissection of molecular and histological subtypes of papillary thyroid cancer using alternative splicing profiles

Despite growing evidence of the relevance of alternative splicing (AS) to cancer development and progression, the biological implications of AS for tumor behaviors, including papillary thyroid cancer (PTC), remain elusive. With the aim of further understanding the molecular and histological subtypes of PTC, we in this study explored whether AS events might act as new molecular determinants. For this purpose, AS profiles were analyzed in RNA-sequencing data from The Cancer Genome Atlas (TCGA) and from a Korean patient dataset. A total of 23 distinct exon-skipping (ES) events that correlated significantly with PTC oncogenic activity and differentiation scores were identified. The two top-ranked ES events, NUMA1_17515 in exon 18 of NUMA1 and TUBB3_38175 in exon 6 of TUBB3, showed high correlations with oncogenic activities and discriminated histological and molecular subtypes of PTC. Furthermore, two novel intron-retention (IR) events for TUBB3 were uncovered. All ES and IR events for the TUBB3 gene were predicted to induce nonsense-mediated mRNA decay. The relative abundances of intron reads in the PTC dataset from TCGA showed IR levels to differ significantly among PTC subtypes, possibly reflecting their different tumor behaviors. This study provides a landscape of AS changes among PTC subtypes and identified two significant AS events, NUMA1_17515 and TUBB3_38175, as potential AS biomarkers for PTC subclassification and characterization. The AS events identified in this study may be involved in the development of phenotypic differences underlying the functional characteristics and histological differentiation of PTCs.

Two potential biomarkers uncovered by scientists in South Korea may help more accurately classify subtypes of papillary thyroid cancer, the most common form of thyroid cancer, and improve treatment regimens. Ascertaining the correct papillary thyroid cancer (PTC) subtype is important for patient prognoses and treatment plans. Growing evidence suggests that cancer progression may be influenced by ‘alternative splicing’ events, alterations to mRNA that change the structure of mRNA transcripts and affect the function of encoded proteins. Yeun-Jun Chung and Sug Hyung Lee at the Catholic University of Korea, Seoul, and co-workers explored alternative splicing events in PTC patient samples. They identified 25 distinct events associated with oncogenic activity and differentiation between PTC subtypes. Of these, two events associated with two separate genes are particularly significant and could prove useful as biomarkers for disease classification and characterisation.

Molecular Pathology of Lung Cancer

This overview of the molecular pathology of lung cancer includes a review of the most salient molecular alterations of the genome, transcriptome, and the epigenome. The insights provided by the growing use of next-generation sequencing (NGS) in lung cancer will be discussed, and interrelated concepts such as intertumor heterogeneity, intratumor heterogeneity, tumor mutational burden, and the advent of liquid biopsy will be explored. Moreover, this work describes how the evolving field of molecular pathology refines the understanding of different histologic phenotypes of non-small-cell lung cancer (NSCLC) and the underlying biology of small-cell lung cancer. This review will provide an appreciation for how ongoing scientific findings and technologic advances in molecular pathology are crucial for development of biomarkers, therapeutic agents, clinical trials, and ultimately improved patient care.

RAS and beyond: the many faces of the neurofibromatosis type 1 protein

Neurofibromatosis type 1 is a rare neurogenetic syndrome, characterized by pigmentary abnormalities, learning and social deficits, and a predisposition for benign and malignant tumor formation caused by germline mutations in the NF1 gene. With the cloning of the NF1 gene and the recognition that the encoded protein, neurofibromin, largely functions as a negative regulator of RAS activity, attention has mainly focused on RAS and canonical RAS effector pathway signaling relevant to disease pathogenesis and treatment. However, as neurofibromin is a large cytoplasmic protein the RAS regulatory domain of which occupies only 10% of its entire coding sequence, both canonical and non-canonical RAS pathway modulation, as well as the existence of potential non-RAS functions, are becoming apparent. In this Special article, we discuss our current understanding of neurofibromin function.

Pathogenic signaling in multiple myeloma

Multiple myeloma is a common hematological malignancy of plasma cells, the terminally differentiated B cells that secrete antibodies as part of the adaptive immune response. Significant progress has been made in treating multiple myeloma, but this disease remains largely incurable, and most patients will eventually suffer a relapse of disease that becomes refractory to further therapies. Moreover, a portion of patients with multiple myeloma present with disease that is refractory to all treatments from the initial diagnosis, and no current therapeutic approaches can help. Therefore, the task remains to advance new therapeutic strategies to help these vulnerable patients. One strategy to meet this challenge is to unravel the complex web of pathogenic signaling pathways in malignant plasma cells and use this information to design novel precision medicine strategies to assist these patients most at risk.

Application of Artificial Intelligence Modeling Technology Based on Fluid Biopsy to Diagnose Alzheimer's Disease

Background: There are no obvious clinical signs and symptoms in the early stages of Alzheimer's disease (AD), and most patients usually have mild cognitive impairment (MCI) before diagnosis. Therefore, early diagnosis of AD is very critical. This paper mainly discusses the blood biomarkers of AD patients and uses machine learning methods to study the changes of blood transcriptome during the development of AD and to search for potential blood biomarkers for AD. Methods: Individualized blood mRNA expression data of 711 patients were downloaded from the GEO database, including the control group (CON) (238 patients), MCI (189 patients), and AD (284 patients). Firstly, we analyzed the subcellular localization, protein types and enrichment pathways of the differentially expressed mRNAs in each group, and established an artificial intelligence individualized diagnostic model. Furthermore, the XCell tool was used to analyze the blood mRNA expression data and obtain blood cell composition and quantitative data. Ratio characteristics were established for mRNA and XCell data. Feature engineering operations such as collinearity and importance analysis were performed on all features to obtain the best feature solicitation. Finally, four machine learning algorithms, including linear support vector machine (SVM), Adaboost, random forest and artificial neural network, were used to model the optimal feature combinations and evaluate their classification performance in the test set. Results: Through feature engineering screening, the best feature collection was obtained. Moreover, the artificial intelligence individualized diagnosis model established based on this method achieved a classification accuracy of 91.59% in the test set. The area under curve (AUC) of CON, MCI, and AD were 0.9746, 0.9536, and 0.9807, respectively. Conclusion: The results of cell homeostasis analysis suggested that the homeostasis of Natural killer T cell (NKT) might be related to AD, and the homeostasis of Granulocyte macrophage progenitor (GMP) might be one of the reasons for AD.

Untangling the clinicopathological significance of MRE11-RAD50-NBS1 complex in sporadic breast cancers

The MRE11-RAD50-NBS1 (MRN) complex is critical for genomic stability. Although germline mutations in MRN may increase breast cancer susceptibility, such mutations are extremely rare. Here, we have conducted a comprehensive clinicopathological study of MRN in sporadic breast cancers. We have protein expression profiled for MRN and a panel of DNA repair factors involved in double-strand break repair (BRCA1, BRCA2, ATM, CHK2, ATR, Chk1, pChk1, RAD51, γH2AX, RPA1, RPA2, DNA-PKcs), RECQ DNA helicases (BLM, WRN, RECQ1, RECQL4, RECQ5), nucleotide excision repair (ERCC1) and base excision repair (SMUG1, APE1, FEN1, PARP1, XRCC1, Pol β) in 1650 clinical breast cancers. The prognostic significance of MRE11, RAD50 and NBS1 transcripts and their microRNA regulators (hsa-miR-494 and hsa-miR-99b) were evaluated in large clinical datasets. Expression of MRN components was analysed in The Cancer Genome Atlas breast cancer cohort. We show that low nuclear MRN is linked to aggressive histopathological phenotypes such as high tumour grade, high mitotic index, oestrogen receptor- and high-risk Nottingham Prognostic Index. In univariate analysis, low nuclear MRE11 and low nuclear RAD50 were associated with poor survival. In multivariate analysis, low nuclear RAD50 remained independently linked with adverse clinical outcomes. Low RAD50 transcripts were also linked with reduced survival. In contrast, overexpression of hsa-miR-494 and hsa-miR-99b microRNAs was associated with poor survival. We observed large-scale genome-wide alterations in MRN-deficient tumours contributing to aggressive behaviour. We conclude that MRN status may be a useful tool to stratify tumours for precision medicine strategies.

Framework for classification of cancer gene expression data using Bayesian hyper-parameter optimization

Computational classification of cancers is an important research problem. Gene expression data has 1000s of features, very few samples, and a class imbalance problem. In this paper, we have proposed a framework for the classification of cancer gene expression profiles. The framework consists of a pipeline of methods for data pre-processing, feature selection, and classification. Data pre-processing is done by standard scaling and normalization of the features. The feature selection is performed in two steps. First, recursive feature elimination (RFE) is used; then, a genetic algorithm is applied only in case RFE results in a feature subset of size more than a specific threshold. Next, is a meta-pool of diverse, individual as well as ensemble classifiers. Hyper-parameters of each member in the meta-pool are optimized using Bayesian Optimization. An algorithm is developed to select the best classifier from the meta-pool based on classification accuracy and computation time taken. We evaluated the framework on 6 publicly available microarray datasets and the PAN-Cancer RNA Sequencing dataset. We found that the classifier selected by the proposed framework produced significant improvement in classification accuracy and computation time required to predict labels for test datasets. A detailed comparison with the state-of-the-art methods shows that the proposed framework outperforms all of them.

Deep Modeling of Regulating Effects of Small Molecules on Longevity-Associated Genes

Aging is considered an inevitable process that causes deleterious effects in the functioning and appearance of cells, tissues, and organs. Recent emergence of large-scale gene expression datasets and significant advances in machine learning techniques have enabled drug repurposing efforts in promoting longevity. In this work, we further developed our previous approach-DeepCOP, a quantitative chemogenomic model that predicts gene regulating effects, and extended its application across multiple cell lines presented in LINCS to predict aging gene regulating effects induced by small molecules. As a result, a quantitative chemogenomic Deep Model was trained using gene ontology labels, molecular fingerprints, and cell line descriptors to predict gene expression responses to chemical perturbations. Other state-of-the-art machine learning approaches were also evaluated as benchmarks. Among those, the deep neural network (DNN) classifier has top-ranked known drugs with beneficial effects on aging genes, and some of these drugs were previously shown to promote longevity, illustrating the potential utility of this methodology. These results further demonstrate the capability of "hybrid" chemogenomic models, incorporating quantitative descriptors from biomarkers to capture cell specific drug-gene interactions. Such models can therefore be used for discovering drugs with desired gene regulatory effects associated with longevity.

Pan-Cancer Analysis Identified C1ORF112 as a Potential Biomarker for Multiple Tumor Types

C1ORF112 is an evolutionarily conserved gene across vertebrates. Over the last decade, studies have suggested that C1ORF112 may play a role in tumorigenesis. Using The Cancer Genome Atlas datasets, we explored the role of C1ORF112 across various tumor types in this study. In most tumor types, C1ORF112 expression was increased in tumor tissues compared to corresponding non-tumor tissues. In patients with certain tumor types, higher C1ORF112 expression was correlated with shorter overall survival, disease-free survival, and progression-free survival. Further analyses of C1ORF112 genetic alteration data showed that C1ORF112 amplification and mutations may have an impact on liver hepatocellular carcinoma and uterine corpus endometrial carcinoma prognosis. In cancers including lower grade glioma and adrenocortical carcinoma, C1ORF112 expression was linked to cancer-associated fibroblast infiltration. Gene Ontology analysis showed that C1ORF112 was co-expressed with genes involved in biological processes such as cell cycle and mitotic regulation. The protein interaction network demonstrated that C1ORF112 physically interacted with RAD51, DMC1, and FIGNL1, which have well characterized functions in DNA repair and cell cycle regulation. This pan-cancer study revealed the prognostic value and oncogenic role of C1ORF112 across multiple tumor types.

Precision omics data integration and analysis with interoperable ontologies and their application for COVID-19 research

Omics technologies are widely used in biomedical research. Precision medicine focuses on individual-level disease treatment and prevention. Here, we propose the usage of the term 'precision omics' to represent the combinatorial strategy that applies omics to translate large-scale molecular omics data for precision disease understanding and accurate disease diagnosis, treatment and prevention. Given the complexity of both omics and precision medicine, precision omics requires standardized representation and integration of heterogeneous data types. Ontology has emerged as an important artificial intelligence component to become critical for standard data and metadata representation, standardization and integration. To support precision omics, we propose a precision omics ontology hypothesis, which hypothesizes that the effectiveness of precision omics is positively correlated with the interoperability of ontologies used for data and knowledge integration. Therefore, to make effective precision omics studies, interoperable ontologies are required to standardize and incorporate heterogeneous data and knowledge in a human- and computer-interpretable manner. Methods for efficient development and application of interoperable ontologies are proposed and illustrated. With the interoperable omics data and knowledge, omics tools such as OmicsViz can also be evolved to process, integrate, visualize and analyze various omics data, leading to the identification of new knowledge and hypotheses of molecular mechanisms underlying the outcomes of diseases such as COVID-19. Given extensive COVID-19 omics research, we propose the strategy of precision omics supported by interoperable ontologies, accompanied with ontology-based semantic reasoning and machine learning, leading to systematic disease mechanism understanding and rational design of precision treatment and prevention.

SHORT ABSTRACT

Precision medicine focuses on individual-level disease treatment and prevention. Precision omics is a new strategy that applies omics for precision medicine research, which requires standardized representation and integration of individual genetics and phenotypes, experimental conditions, and data analysis settings. Ontology has emerged as an important artificial intelligence component to become critical for standard data and metadata representation, standardization and integration. To support precision omics, interoperable ontologies are required in order to standardize and incorporate heterogeneous data and knowledge in a human- and computer-interpretable manner. With the interoperable omics data and knowledge, omics tools such as OmicsViz can also be evolved to process, integrate, visualize and analyze various omics data, leading to the identification of new knowledge and hypotheses of molecular mechanisms underlying disease outcomes. The precision COVID-19 omics study is provided as the primary use case to illustrate the rationale and implementation of the precision omics strategy.

Machine learning analysis of TCGA cancer data

In recent years, machine learning (ML) researchers have changed their focus towards biological problems that are difficult to analyse with standard approaches. Large initiatives such as The Cancer Genome Atlas (TCGA) have allowed the use of omic data for the training of these algorithms. In order to study the state of the art, this review is provided to cover the main works that have used ML with TCGA data. Firstly, the principal discoveries made by the TCGA consortium are presented. Once these bases have been established, we begin with the main objective of this study, the identification and discussion of those works that have used the TCGA data for the training of different ML approaches. After a review of more than 100 different papers, it has been possible to make a classification according to following three pillars: the type of tumour, the type of algorithm and the predicted biological problem. One of the conclusions drawn in this work shows a high density of studies based on two major algorithms: Random Forest and Support Vector Machines. We also observe the rise in the use of deep artificial neural networks. It is worth emphasizing, the increase of integrative models of multi-omic data analysis. The different biological conditions are a consequence of molecular homeostasis, driven by both protein coding regions, regulatory elements and the surrounding environment. It is notable that a large number of works make use of genetic expression data, which has been found to be the preferred method by researchers when training the different models. The biological problems addressed have been classified into five types: prognosis prediction, tumour subtypes, microsatellite instability (MSI), immunological aspects and certain pathways of interest. A clear trend was detected in the prediction of these conditions according to the type of tumour. That is the reason for which a greater number of works have focused on the BRCA cohort, while specific works for survival, for example, were centred on the GBM cohort, due to its large number of events. Throughout this review, it will be possible to go in depth into the works and the methodologies used to study TCGA cancer data. Finally, it is intended that this work will serve as a basis for future research in this field of study.

Targeting acute myeloid leukemia dependency on VCP-mediated DNA repair through a selective second-generation small-molecule inhibitor

The development and survival of cancer cells require adaptive mechanisms to stress. Such adaptations can confer intrinsic vulnerabilities, enabling the selective targeting of cancer cells. Through a pooled in vivo short hairpin RNA (shRNA) screen, we identified the adenosine triphosphatase associated with diverse cellular activities (AAA-ATPase) valosin-containing protein (VCP) as a top stress-related vulnerability in acute myeloid leukemia (AML). We established that AML was the most responsive disease to chemical inhibition of VCP across a panel of 16 cancer types. The sensitivity to VCP inhibition of human AML cell lines, primary patient samples, and syngeneic and xenograft mouse models of AML was validated using VCP-directed shRNAs, overexpression of a dominant-negative VCP mutant, and chemical inhibition. By combining mass spectrometry-based analysis of the VCP interactome and phospho-signaling studies, we determined that VCP is important for ataxia telangiectasia mutated (ATM) kinase activation and subsequent DNA repair through homologous recombination in AML. A second-generation VCP inhibitor, CB-5339, was then developed and characterized. Efficacy and safety of CB-5339 were validated in multiple AML models, including syngeneic and patient-derived xenograft murine models. We further demonstrated that combining DNA-damaging agents, such as anthracyclines, with CB-5339 treatment synergizes to impair leukemic growth in an MLL-AF9-driven AML murine model. These studies support the clinical testing of CB-5339 as a single agent or in combination with standard-of-care DNA-damaging chemotherapy for the treatment of AML.

Making Sense of Genetic Information: The Promising Evolution of Clinical Stratification and Precision Oncology Using Machine Learning

Precision medicine is a medical approach to administer patients with a tailored dose of treatment by taking into consideration a person's variability in genes, environment, and lifestyles. The accumulation of omics big sequence data led to the development of various genetic databases on which clinical stratification of high-risk populations may be conducted. In addition, because cancers are generally caused by tumor-specific mutations, large-scale systematic identification of single nucleotide polymorphisms (SNPs) in various tumors has propelled significant progress of tailored treatments of tumors (i.e., precision oncology). Machine learning (ML), a subfield of artificial intelligence in which computers learn through experience, has a great potential to be used in precision oncology chiefly to help physicians make diagnostic decisions based on tumor images. A promising venue of ML in precision oncology is the integration of all available data from images to multi-omics big data for the holistic care of patients and high-risk healthy subjects. In this review, we provide a focused overview of precision oncology and ML with attention to breast cancer and glioma as well as the Bayesian networks that have the flexibility and the ability to work with incomplete information. We also introduce some state-of-the-art attempts to use and incorporate ML and genetic information in precision oncology.

Systematic interrogation of mutation groupings reveals divergent downstream expression programs within key cancer genes

Background

Genes implicated in tumorigenesis often exhibit diverse sets of genomic variants in the tumor cohorts within which they are frequently mutated. For many genes, neither the transcriptomic effects of these variants nor their relationship to one another in cancer processes have been well-characterized. We sought to identify the downstream expression effects of these mutations and to determine whether this heterogeneity at the genomic level is reflected in a corresponding heterogeneity at the transcriptomic level.

Results

By applying a novel hierarchical framework for organizing the mutations present in a cohort along with machine learning pipelines trained on samples’ expression profiles we systematically interrogated the signatures associated with combinations of mutations recurrent in cancer. This allowed us to catalogue the mutations with discernible downstream expression effects across a number of tumor cohorts as well as to uncover and characterize over a hundred cases where subsets of a gene’s mutations are clearly divergent in their function from the remaining mutations of the gene. These findings successfully replicated across a number of disease contexts and were found to have clear implications for the delineation of cancer processes and for clinical decisions.

Conclusions

The results of cataloguing the downstream effects of mutation subgroupings across cancer cohorts underline the importance of incorporating the diversity present within oncogenes in models designed to capture the downstream effects of their mutations.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12859-021-04147-y.

Building patient-specific models for receptor tyrosine kinase signaling networks

Cancer progresses due to changes in the dynamic interactions of multidimensional factors associated with gene mutations. Cancer research has actively adopted computational methods, including data-driven and mathematical model-driven approaches, to identify causative factors and regulatory rules that can explain the complexity and diversity of cancers. A data-driven, statistics-based approach revealed correlations between gene alterations and clinical outcomes in many types of cancers. A model-driven mathematical approach has elucidated the dynamic features of cancer networks and identified the mechanisms of drug efficacy and resistance. More recently, machine learning methods have emerged that can be used for mining omics data and classifying patient. However, as the strengths and weaknesses of each method becoming apparent, new analytical tools are emerging to combine and improve the methodologies and maximize their predictive power for classifying cancer subtypes and prognosis. Here, we introduce recent advances in cancer systems biology aimed at personalized medicine, with focus on the receptor tyrosine kinase signaling network.

Incorporating Machine Learning into Established Bioinformatics Frameworks

The exponential growth of biomedical data in recent years has urged the application of numerous machine learning techniques to address emerging problems in biology and clinical research. By enabling the automatic feature extraction, selection, and generation of predictive models, these methods can be used to efficiently study complex biological systems. Machine learning techniques are frequently integrated with bioinformatic methods, as well as curated databases and biological networks, to enhance training and validation, identify the best interpretable features, and enable feature and model investigation. Here, we review recently developed methods that incorporate machine learning within the same framework with techniques from molecular evolution, protein structure analysis, systems biology, and disease genomics. We outline the challenges posed for machine learning, and, in particular, deep learning in biomedicine, and suggest unique opportunities for machine learning techniques integrated with established bioinformatics approaches to overcome some of these challenges.

Super enhancers define regulatory subtypes and cell identity in neuroblastoma

Half of the children diagnosed with neuroblastoma (NB) have high-risk disease, disproportionately contributing to overall childhood cancer-related deaths. In addition to recurrent gene mutations, there is increasing evidence supporting the role of epigenetic deregulation in disease pathogenesis. Yet, comprehensive cis-regulatory network descriptions from NB are lacking. Here, using genome-wide H3K27ac profiles across 60 NBs, covering the different clinical and molecular subtypes, we identified four major super-enhancer-driven epigenetic subtypes and their underlying master regulatory networks. Three of these subtypes recapitulated known clinical groups; namely, MYCN-amplified, MYCN non-amplified high-risk and MYCN non-amplified low-risk NBs. The fourth subtype, exhibiting mesenchymal characteristics, shared cellular identity with multipotent Schwann cell precursors, was induced by RAS activation and was enriched in relapsed disease. Notably, CCND1, an essential gene in NB, was regulated by both mesenchymal and adrenergic regulatory networks converging on distinct super-enhancer modules. Overall, this study reveals subtype-specific super-enhancer regulation in NBs.

Prediction of PIK3CA mutations from cancer gene expression data

Breast cancers with PIK3CA mutations can be treated with PIK3CA inhibitors in hormone receptor-positive HER2 negative subtypes. We applied a supervised elastic net penalized logistic regression model to predict PIK3CA mutations from gene expression data. This regression approach was applied to predict modeling using the TCGA pan-cancer dataset. Approximately 10,000 cases were available for PIK3CA mutation and mRNA expression data. In 10-fold cross-validation, the model with λ = 0.01 and α = 1.0 (ridge regression) showed the best performance, in terms of area under the receiver operating characteristic (AUROC). The final model was developed with selected hyper-parameters using the entire training set. The training set AUROC was 0.93, and the test set AUROC was 0.84. The area under the precision-recall (AUPR) of the training set was 0.66, and the test set AUPR was 0.39. Cancer types were the most important predictors. Both insulin like growth factor 1 receptor (IGF1R) and the phosphatase and tensin homolog (PTEN) were the most significant genes in gene expression predictors. Our study suggests that predicting genomic alterations using gene expression data is possible, with good outcomes.

Identification of pan-cancer Ras pathway activation with deep learning

The identification of hidden responders is often an essential challenge in precision oncology. A recent attempt based on machine learning has been proposed for classifying aberrant pathway activity from multiomic cancer data. However, we note several critical limitations there, such as high-dimensionality, data sparsity and model performance. Given the central importance and broad impact of precision oncology, we propose nature-inspired deep Ras activation pan-cancer (NatDRAP), a deep neural network (DNN) model, to address those restrictions for the identification of hidden responders. In this study, we develop the nature-inspired deep learning model that integrates bulk RNA sequencing, copy number and mutation data from PanCanAltas to detect pan-cancer Ras pathway activation. In NatDRAP, we propose to synergize the nature-inspired artificial bee colony algorithm with different gradient-based optimizers in one framework for optimizing DNNs in a collaborative manner. Multiple experiments were conducted on 33 different cancer types across PanCanAtlas. The experimental results demonstrate that the proposed NatDRAP can provide superior performance over other benchmark methods with strong robustness towards diagnosing RAS aberrant pathway activity across different cancer types. In addition, gene ontology enrichment and pathological analysis are conducted to reveal novel insights into the RAS aberrant pathway activity identification and characterization. NatDRAP is written in Python and available at https://github.com/lixt314/NatDRAP1.

Cutting the Brakes on Ras-Cytoplasmic GAPs as Targets of Inactivation in Cancer

Simple Summary

GTPase-Activating Proteins (RasGAPs) are a group of structurally related proteins with a fundamental role in controlling the activity of Ras in normal and cancer cells. In particular, loss of function of RasGAPs may contribute to aberrant Ras activation in cancer. Here we review the multiple molecular mechanisms and factors that are involved in downregulating RasGAPs expression and functions in cancer. Additionally, we discuss how extracellular stimuli from the tumor microenvironment can control RasGAPs expression and activity in cancer cells and stromal cells, indirectly affecting Ras activation, with implications for cancer development and progression.

Abstract

The Ras pathway is frequently deregulated in cancer, actively contributing to tumor development and progression. Oncogenic activation of the Ras pathway is commonly due to point mutation of one of the three Ras genes, which occurs in almost one third of human cancers. In the absence of Ras mutation, the pathway is frequently activated by alternative means, including the loss of function of Ras inhibitors. Among Ras inhibitors, the GTPase-Activating Proteins (RasGAPs) are major players, given their ability to modulate multiple cancer-related pathways. In fact, most RasGAPs also have a multi-domain structure that allows them to act as scaffold or adaptor proteins, affecting additional oncogenic cascades. In cancer cells, various mechanisms can cause the loss of function of Ras inhibitors; here, we review the available evidence of RasGAP inactivation in cancer, with a specific focus on the mechanisms. We also consider extracellular inputs that can affect RasGAP levels and functions, implicating that specific conditions in the tumor microenvironment can foster or counteract Ras signaling through negative or positive modulation of RasGAPs. A better understanding of these conditions might have relevant clinical repercussions, since treatments to restore or enhance the function of RasGAPs in cancer would help circumvent the intrinsic difficulty of directly targeting the Ras protein.

Deep transfer learning for reducing health care disparities arising from biomedical data inequality

As artificial intelligence (AI) is increasingly applied to biomedical research and clinical decisions, developing unbiased AI models that work equally well for all ethnic groups is of crucial importance to health disparity prevention and reduction. However, the biomedical data inequality between different ethnic groups is set to generate new health care disparities through data-driven, algorithm-based biomedical research and clinical decisions. Using an extensive set of machine learning experiments on cancer omics data, we find that current prevalent schemes of multiethnic machine learning are prone to generating significant model performance disparities between ethnic groups. We show that these performance disparities are caused by data inequality and data distribution discrepancies between ethnic groups. We also find that transfer learning can improve machine learning model performance for data-disadvantaged ethnic groups, and thus provides an effective approach to reduce health care disparities arising from data inequality among ethnic groups.

SOS GEFs in health and disease

SOS1 and SOS2 are the most universal and widely expressed family of guanine exchange factors (GEFs) capable or activating RAS or RAC1 proteins in metazoan cells. SOS proteins contain a sequence of modular domains that are responsible for different intramolecular and intermolecular interactions modulating mechanisms of self-inhibition, allosteric activation and intracellular homeostasis. Despite their homology, analyses of SOS1/2-KO mice demonstrate functional prevalence of SOS1 over SOS2 in cellular processes including proliferation, migration, inflammation or maintenance of intracellular redox homeostasis, although some functional redundancy cannot be excluded, particularly at the organismal level. Specific SOS1 gain-of-function mutations have been identified in inherited RASopathies and various sporadic human cancers. SOS1 depletion reduces tumorigenesis mediated by RAS or RAC1 in mouse models and is associated with increased intracellular oxidative stress and mitochondrial dysfunction. Since WT RAS is essential for development of RAS-mutant tumors, the SOS GEFs may be considered as relevant biomarkers or therapy targets in RAS-dependent cancers. Inhibitors blocking SOS expression, intrinsic GEF activity, or productive SOS protein-protein interactions with cellular regulators and/or RAS/RAC targets have been recently developed and shown preclinical and clinical effectiveness blocking aberrant RAS signaling in RAS-driven and RTK-driven tumors.