Deep convolutional neural networks for accurate somatic mutation detection

Comprehensive identification of pathogenic variants in retinoblastoma by long- and short-read sequencing

Retinoblastoma (RB) is the most common intraocular malignancy in childhood. The causal variants in RB are mostly characterized by previously used short-read sequencing (SRS) analysis, which has technical limitations in identifying structural variants (SVs) and phasing information. Long-read sequencing (LRS) technology has advantages over SRS in detecting SVs, phased genetic variants, and methylation. In this study, we comprehensively characterized the genetic landscape of RB using combinatorial LRS and SRS of 16 RB tumors and 16 matched blood samples. We detected a total of 232 somatic SVs, with an average of 14.5 SVs per sample across the whole genome in our cohort. We identified 20 distinct pathogenic variants disrupting RB1 gene, including three novel small variants and five somatic SVs. We found more somatic SVs were detected from LRS than SRS (140 vs. 122) in RB samples with WGS data, particularly the insertions (18 vs. 1). Furthermore, our analysis shows that, with the exception of one sample who lacked the methylation data, all samples presented biallelic inactivation of RB1 in various forms, including two cases with the biallelic hypermethylated promoter and four cases with compound heterozygous mutations which were missing in SRS analysis. By inferring relative timing of somatic events, we reveal the genetic progression that RB1 disruption early and followed by copy number changes, including amplifications of Chr2p and deletions of Chr16q, during RB tumorigenesis. Altogether, we characterize the comprehensive genetic landscape of RB, providing novel insights into the genetic alterations and mechanisms contributing to RB initiation and development. Our work also establishes a framework to analyze genomic landscape of cancers based on LRS data.

Bioinformatics tools and resources for cancer and application

ABSTRACT

Tumor bioinformatics plays an important role in cancer research and precision medicine. The primary focus of traditional cancer research has been molecular and clinical studies of a number of fundamental pathways and genes. In recent years, driven by breakthroughs in high-throughput technologies, large-scale cancer omics data have accumulated rapidly. How to effectively utilize and share these data is particularly important. To address this crucial task, many computational tools and databases have been developed over the past few years. To help researchers quickly learn and understand the functions of these tools, in this review, we summarize publicly available bioinformatics tools and resources for pan-cancer multi-omics analysis, regulatory analysis of tumorigenesis, tumor treatment and prognosis, immune infiltration analysis, immune repertoire analysis, cancer driver gene and driver mutation analysis, and cancer single-cell analysis, which may further help researchers find more suitable tools for their research.

A Novel Affordable and Reliable Framework for Accurate Detection and Comprehensive Analysis of Somatic Mutations in Cancer

Accurate detection and analysis of somatic variants in cancer involve multiple third-party tools with complex dependencies and configurations, leading to laborious, error-prone, and time-consuming data conversions. This approach lacks accuracy, reproducibility, and portability, limiting clinical application. Musta was developed to address these issues as an end-to-end pipeline for detecting, classifying, and interpreting cancer mutations. Musta is based on a Python command-line tool designed to manage tumor-normal samples for precise somatic mutation analysis. The core is a Snakemake-based workflow that covers all key cancer genomics steps, including variant calling, mutational signature deconvolution, variant annotation, driver gene detection, pathway analysis, and tumor heterogeneity estimation. Musta is easy to install on any system via Docker, with a Makefile handling installation, configuration, and execution, allowing for full or partial pipeline runs. Musta has been validated at the CRS4-NGS Core facility and tested on large datasets from The Cancer Genome Atlas and the Beijing Institute of Genomics. Musta has proven robust and flexible for somatic variant analysis in cancer. It is user-friendly, requiring no specialized programming skills, and enables data processing with a single command line. Its reproducibility ensures consistent results across users following the same protocol.

Genomic Mosaicism of the Brain: Origin, Impact, and Utility

Genomic mosaicism describes the phenomenon where some but not all cells within a tissue harbor unique genetic mutations. Traditionally, research focused on the impact of genomic mosaicism on clinical phenotype-motivated by its involvement in cancers and overgrowth syndromes. More recently, we increasingly shifted towards the plethora of neutral mosaic variants that can act as recorders of cellular lineage and environmental exposures. Here, we summarize the current state of the field of genomic mosaicism research with a special emphasis on our current understanding of this phenomenon in brain development and homeostasis. Although the field of genomic mosaicism has a rich history, technological advances in the last decade have changed our approaches and greatly improved our knowledge. We will provide current definitions and an overview of contemporary detection approaches for genomic mosaicism. Finally, we will discuss the impact and utility of genomic mosaicism.

Overcoming Limitations to Deep Learning in Domesticated Animals with TrioTrain

Variant calling across diverse species remains challenging as most bioinformatics tools default to assumptions based on human genomes. DeepVariant (DV) excels without joint genotyping while offering fewer implementation barriers. However, the growing appeal of a "universal" algorithm has magnified the unknown impacts when used with non-human genomes. Here, we use bovine genomes to assess the limits of human-genome-trained models in other species. We introduce the first multi-species DV model that achieves a lower Mendelian Inheritance Error (MIE) rate during single-sample genotyping. Our novel approach, TrioTrain, automates extending DV for species without Genome In A Bottle (GIAB) resources and uses region shuffling to mitigate barriers for SLURM-based clusters. To offset imperfect truth labels for animal genomes, we remove Mendelian discordant variants before training, where models are tuned to genotype the offspring correctly. With TrioTrain, we use cattle, yak, and bison trios to build 30 model iterations across five phases. We observe remarkable performance across phases when testing the GIAB human trios with a mean SNP F1 score >0.990. In HG002, our phase 4 bovine model identifies more variants at a lower MIE rate than DeepTrio. In bovine F1-hybrid genomes, our model substantially reduces inheritance errors with a mean MIE rate of 0.03 percent. Although constrained by imperfect labels, we find that multi-species, trio-based training produces a robust variant calling model. Our research demonstrates that exclusively training with human genomes restricts the application of deep-learning approaches for comparative genomics.

COSAP: Comparative Sequencing Analysis Platform

BACKGROUND

Recent improvements in sequencing technologies enabled detailed profiling of genomic features. These technologies mostly rely on short reads which are merged and compared to reference genome for variant identification. These operations should be done with computers due to the size and complexity of the data. The need for analysis software resulted in many programs for mapping, variant calling and annotation steps. Currently, most programs are either expensive enterprise software with proprietary code which makes access and verification very difficult or open-access programs that are mostly based on command-line operations without user interfaces and extensive documentation. Moreover, a high level of disagreement is observed among popular mapping and variant calling algorithms in multiple studies, which makes relying on a single algorithm unreliable. User-friendly open-source software tools that offer comparative analysis are an important need considering the growth of sequencing technologies.

RESULTS

Here, we propose Comparative Sequencing Analysis Platform (COSAP), an open-source platform that provides popular sequencing algorithms for SNV, indel, structural variant calling, copy number variation, microsatellite instability and fusion analysis and their annotations. COSAP is packed with a fully functional user-friendly web interface and a backend server which allows full independent deployment for both individual and institutional scales. COSAP is developed as a workflow management system and designed to enhance cooperation among scientists with different backgrounds. It is publicly available at https://cosap.bio and https://github.com/MBaysanLab/cosap/ . The source code of the frontend and backend services can be found at https://github.com/MBaysanLab/cosap-webapi/ and https://github.com/MBaysanLab/cosap_frontend/ respectively. All services are packed as Docker containers as well. Pipelines that combine algorithms can be customized and new algorithms can be added with minimal coding through modular structure.

CONCLUSIONS

COSAP simplifies and speeds up the process of DNA sequencing analyses providing commonly used algorithms for SNV, indel, structural variant calling, copy number variation, microsatellite instability and fusion analysis as well as their annotations. COSAP is packed with a fully functional user-friendly web interface and a backend server which allows full independent deployment for both individual and institutional scales. Standardized implementations of popular algorithms in a modular platform make comparisons much easier to assess the impact of alternative pipelines which is crucial in establishing reproducibility of sequencing analyses.

Advances in Machine Learning Processing of Big Data from Disease Diagnosis Sensors

Exploring accurate, noninvasive, and inexpensive disease diagnostic sensors is a critical task in the fields of chemistry, biology, and medicine. The complexity of biological systems and the explosive growth of biomarker data have driven machine learning to become a powerful tool for mining and processing big data from disease diagnosis sensors. With the development of bioinformatics and artificial intelligence (AI), machine learning models formed by data mining have been able to guide more sensitive and accurate molecular computing. This review presents an overview of big data collection approaches and fundamental machine learning algorithms and discusses recent advances in machine learning and molecular computational disease diagnostic sensors. More specifically, we highlight existing modular workflows and key opportunities and challenges for machine learning to achieve disease diagnosis through big data mining.

Improving somatic exome sequencing performance by biological replicates

BACKGROUND

Next-generation sequencing (NGS) technologies offer fast and inexpensive identification of DNA sequences. Somatic sequencing is among the primary applications of NGS, where acquired (non-inherited) variants are based on comparing diseased and healthy tissues from the same individual. Somatic mutations in genetic diseases such as cancer are tightly associated with genomic instability. Genomic instability increases heterogenity, complicating sequencing efforts further, a task already challenged by the presence of short reads and repetitions in human DNA. This leads to low concordance among studies and limits reproducibility. This limitation is a significant problem since identified mutations in somatic sequencing are major biomarkers for diagnosis and the primary input of targeted therapies. Benchmarking studies were conducted to assess the error rates and increase reproducibility. Unfortunately, the number of somatic benchmarking sets is very limited due to difficulties in validating true somatic variants. Moreover, most NGS benchmarking studies are based on relatively simpler germline (inherited) sequencing. Recently, a comprehensive somatic sequencing benchmarking set was published by Sequencing Quality Control Phase 2 (SEQC2). We chose this dataset for our experiments because it is a well-validated, cancer-focused dataset that includes many tumor/normal biological replicates. Our study has two primary goals. First goal is to determine how replicate-based consensus approaches can improve the accuracy of somatic variant detection systems. Second goal is to develop highly predictive machine learning (ML) models by employing replicate-based consensus variants as labels during the training phase.

RESULTS

Ensemble approaches that combine alternative algorithms are relatively common; here, as an alternative, we study the performance enhancement potential of biological replicates. We first developed replicate-based consensus approaches that utilize the biological replicates available in this study to improve variant calling performance. Subsequently, we trained ML models using these biological replicates and achieved performance comparable to optimal ML models, those trained using high-confidence variants identified in advance.

CONCLUSIONS

Our replicate-based consensus approach can be used to improve variant calling performance and develop efficient ML models. Given the relative ease of obtaining biological replicates, this strategy allows for the development of efficient ML models tailored to specific datasets or scenarios.

Emerging applications of machine learning in genomic medicine and healthcare

The integration of artificial intelligence technologies has propelled the progress of clinical and genomic medicine in recent years. The significant increase in computing power has facilitated the ability of artificial intelligence models to analyze and extract features from extensive medical data and images, thereby contributing to the advancement of intelligent diagnostic tools. Artificial intelligence (AI) models have been utilized in the field of personalized medicine to integrate clinical data and genomic information of patients. This integration allows for the identification of customized treatment recommendations, ultimately leading to enhanced patient outcomes. Notwithstanding the notable advancements, the application of artificial intelligence (AI) in the field of medicine is impeded by various obstacles such as the limited availability of clinical and genomic data, the diversity of datasets, ethical implications, and the inconclusive interpretation of AI models' results. In this review, a comprehensive evaluation of multiple machine learning algorithms utilized in the fields of clinical and genomic medicine is conducted. Furthermore, we present an overview of the implementation of artificial intelligence (AI) in the fields of clinical medicine, drug discovery, and genomic medicine. Finally, a number of constraints pertaining to the implementation of artificial intelligence within the healthcare industry are examined.

FCMSTrans: Accurate Prediction of Disease-Associated nsSNPs by Utilizing Multiscale Convolution and Deep Feature Combination within a Transformer Framework

Nonsynonymous single-nucleotide polymorphisms (nsSNPs), implicated in over 6000 diseases, necessitate accurate prediction for expedited drug discovery and improved disease diagnosis. In this study, we propose FCMSTrans, a novel nsSNP predictor that innovatively combines the transformer framework and multiscale modules for comprehensive feature extraction. The distinctive attribute of FCMSTrans resides in a deep feature combination strategy. This strategy amalgamates evolutionary-scale modeling (ESM) and ProtTrans (PT) features, providing an understanding of protein biochemical properties, and position-specific scoring matrix, secondary structure, predicted relative solvent accessibility, and predicted disorder (PSPP) features, which are derived from four protein sequences and structure-oriented characteristics. This feature combination offers a comprehensive view of the molecular dynamics involving nsSNPs. Our model employs the transformer's self-attention mechanisms across multiple layers, extracting higher-level and abstract representations. Simultaneously, varied-level features are captured by multiscale convolutions, enriching feature abstraction at multiple echelons. Our comparative analyses with existing methodologies highlight significant improvements made possible by the integrated feature fusion approach adopted in FCMSTrans. This is further substantiated by performance assessments based on diverse data sets, such as PredictSNP, MMP, and PMD, with areas under the curve (AUCs) of 0.869, 0.819, and 0.693, respectively. Furthermore, FCMSTrans shows robustness and superiority by outperforming the current best predictor, PROVEAN, in a blind test conducted on a third-party data set, achieving an impressive AUC score of 0.7838. The Python code of FCMSTrans is available at https://github.com/gc212/FCMSTrans for academic usage.

The protective role of Achyranthes aspera extract against cisplatin-induced nephrotoxicity by alleviating oxidative stress, inflammation, and PANoptosis

ETHNOPHARMACOLOGICAL RELEVANCE

Achyranthes aspera, a widely recognized medicinal plant, is used in various cultures for treating different ailments, including renal dysfunction; however, there is a lack of comprehensive understanding of its protective effects and the underlying signaling networks involved.

AIM OF THE STUDY

This study aimed to investigate the molecular mechanisms of the action of A. aspera by employing an integrative approach including functional and tissue imaging as well as comprehensive genomics analysis.

MATERIALS AND METHODS

Cisplatin-induced nephrotoxicity is a well-established animal model for acute kidney injury (AKI). In this study, we investigated the protective effects and underlying mechanisms of the action of A. aspera water-soluble extract (AAW) on a murine model of cisplatin-induced AKI. The evaluation includes measurements of blood urea nitrogen (BUN) and serum creatinine (SCr) levels, histology examination, and transcriptome analysis using RNA sequencing.

RESULTS

In male ICR mice, oral administration of AAW at doses of 0.5-1.0 g/kg significantly reduced cisplatin-induced nephrotoxicity. This effect included the amelioration of tubular injury, renal fibrosis, and the lowering of BUN and SCr levels. AAW also effectively decreased oxidative markers, such as malondialdehyde (MDA) and nitrotyrosine (NT), along with inflammation markers, including COX-2, iNOS, NLRP3, and pP65NFκB. Moreover, AAW administration induced a dose-dependent increase in the expression of two protective factors, Nrf2 and BcL2, and suppressed apoptosis, as evidenced by reduced levels of truncated caspase 3 (t-Casp3). To explore the underlying molecular mechanisms and signaling networks, next-generation sequencing (NGS) analysis was employed. The results revealed that AAW mitigated apoptosis, necroptosis, and PANoptosis pathways by inhibiting inflammation signaling pathways, such as the TNFα-, NFκB-, NETs-, and leukocyte transendothelial migration pathways. Additionally, AAW was found to enhance protective signaling pathways, including the cGMP/PKG-, cAMP-, AMPK-, and mTOR-dependent activation of autophagy and mitophagy pathways. The primary bioactive compound found in AAW was identified as 20-hydroxyecdysone (0.36%).

CONCLUSION

Our study demonstrates that AAW reduces cisplatin-induced nephrotoxicity. The protective effects of AAW are attributed to its modulation of multiple molecular signaling networks. Specifically, AAW downregulates genes and signaling pathways associated with oxidative stress and endoplasmic reticulum (ER) stress, inflammation, and PANoptosis. Simultaneously, it upregulates genes and signaling pathways associated with cell survival, including autophagy and mitophagy pathways.

Evaluating the performance of low-frequency variant calling tools for the detection of variants from short-read deep sequencing data

Detection of low-frequency variants with high accuracy plays an important role in biomedical research and clinical practice. However, it is challenging to do so with next-generation sequencing (NGS) approaches due to the high error rates of NGS. To accurately distinguish low-level true variants from these errors, many statistical variants calling tools for calling low-frequency variants have been proposed, but a systematic performance comparison of these tools has not yet been performed. Here, we evaluated four raw-reads-based variant callers (SiNVICT, outLyzer, Pisces, and LoFreq) and four UMI-based variant callers (DeepSNVMiner, MAGERI, smCounter2, and UMI-VarCal) considering their capability to call single nucleotide variants (SNVs) with allelic frequency as low as 0.025% in deep sequencing data. We analyzed a total of 54 simulated data with various sequencing depths and variant allele frequencies (VAFs), two reference data, and Horizon Tru-Q sample data. The results showed that the UMI-based callers, except smCounter2, outperformed the raw-reads-based callers regarding detection limit. Sequencing depth had almost no effect on the UMI-based callers but significantly influenced on the raw-reads-based callers. Regardless of the sequencing depth, MAGERI showed the fastest analysis, while smCounter2 consistently took the longest to finish the variant calling process. Overall, DeepSNVMiner and UMI-VarCal performed the best with considerably good sensitivity and precision of 88%, 100%, and 84%, 100%, respectively. In conclusion, the UMI-based callers, except smCounter2, outperformed the raw-reads-based callers in terms of sensitivity and precision. We recommend using DeepSNVMiner and UMI-VarCal for low-frequency variant detection. The results provide important information regarding future directions for reliable low-frequency variant detection and algorithm development, which is critical in genetics-based medical research and clinical applications.

Neoantigen identification: Technological advances and challenges

Neoantigens have emerged as promising targets for cutting-edge immunotherapies, such as cancer vaccines and adoptive cell therapy. These neoantigens are unique to tumors and arise exclusively from somatic mutations or non-genomic aberrations in tumor proteins. They encompass a wide range of alterations, including genomic mutations, post-transcriptomic variants, and viral oncoproteins. With the advancements in technology, the identification of immunogenic neoantigens has seen rapid progress, raising new opportunities for enhancing their clinical significance. Prediction of neoantigens necessitates the acquisition of high-quality samples and sequencing data, followed by mutation calling. Subsequently, the pipeline involves integrating various tools that can predict the expression, processing, binding, and recognition potential of neoantigens. However, the continuous improvement of computational tools is constrained by the availability of datasets which contain validated immunogenic neoantigens. This review article aims to provide a comprehensive summary of the current knowledge as well as limitations in neoantigen prediction and validation. Additionally, it delves into the origin and biological role of neoantigens, offering a deeper understanding of their significance in the field of cancer immunotherapy. This article thus seeks to contribute to the ongoing efforts to harness neoantigens as powerful weapons in the fight against cancer.

AIVariant: a deep learning-based somatic variant detector for highly contaminated tumor samples

The detection of somatic DNA variants in tumor samples with low tumor purity or sequencing depth remains a daunting challenge despite numerous attempts to address this problem. In this study, we constructed a substantially extended set of actual positive variants originating from a wide range of tumor purities and sequencing depths, as well as actual negative variants derived from sequencer-specific sequencing errors. A deep learning model named AIVariant, trained on this extended dataset, outperforms previously reported methods when tested under various tumor purities and sequencing depths, especially low tumor purity and sequencing depth.

Control-independent mosaic single nucleotide variant detection with DeepMosaic

Mosaic variants (MVs) reflect mutagenic processes during embryonic development and environmental exposure, accumulate with aging and underlie diseases such as cancer and autism. The detection of noncancer MVs has been computationally challenging due to the sparse representation of nonclonally expanded MVs. Here we present DeepMosaic, combining an image-based visualization module for single nucleotide MVs and a convolutional neural network-based classification module for control-independent MV detection. DeepMosaic was trained on 180,000 simulated or experimentally assessed MVs, and was benchmarked on 619,740 simulated MVs and 530 independent biologically tested MVs from 16 genomes and 181 exomes. DeepMosaic achieved higher accuracy compared with existing methods on biological data, with a sensitivity of 0.78, specificity of 0.83 and positive predictive value of 0.96 on noncancer whole-genome sequencing data, as well as doubling the validation rate over previous best-practice methods on noncancer whole-exome sequencing data (0.43 versus 0.18). DeepMosaic represents an accurate MV classifier for noncancer samples that can be implemented as an alternative or complement to existing methods.

Self-Curable Synaptic Ferroelectric FET Arrays for Neuromorphic Convolutional Neural Network

With the recently increasing prevalence of deep learning, both academia and industry exhibit substantial interest in neuromorphic computing, which mimics the functional and structural features of the human brain. To realize neuromorphic computing, an energy-efficient and reliable artificial synapse must be developed. In this study, the synaptic ferroelectric field-effect-transistor (FeFET) array is fabricated as a component of a neuromorphic convolutional neural network. Beyond the single transistor level, the long-term potentiation and depression of synaptic weights are achieved at the array level, and a successful program-inhibiting operation is demonstrated in the synaptic array, achieving a learning accuracy of 79.84% on the Canadian Institute for Advanced Research (CIFAR)-10 dataset. Furthermore, an efficient self-curing method is proposed to improve the endurance of the FeFET array by tenfold, utilizing the punch-through current inherent to the device. Low-frequency noise spectroscopy is employed to quantitatively evaluate the curing efficiency of the proposed self-curing method. The results of this study provide a method to fabricate and operate reliable synaptic FeFET arrays, thereby paving the way for further development of ferroelectric-based neuromorphic computing.

Validation of genetic variants from NGS data using deep convolutional neural networks

Accurate somatic variant calling from next-generation sequencing data is one most important tasks in personalised cancer therapy. The sophistication of the available technologies is ever-increasing, yet, manual candidate refinement is still a necessary step in state-of-the-art processing pipelines. This limits reproducibility and introduces a bottleneck with respect to scalability. We demonstrate that the validation of genetic variants can be improved using a machine learning approach resting on a Convolutional Neural Network, trained using existing human annotation. In contrast to existing approaches, we introduce a way in which contextual data from sequencing tracks can be included into the automated assessment. A rigorous evaluation shows that the resulting model is robust and performs on par with trained researchers following published standard operating procedure.

Variant calling and benchmarking in an era of complete human genome sequences

Genetic variant calling from DNA sequencing has enabled understanding of germline variation in hundreds of thousands of humans. Sequencing technologies and variant-calling methods have advanced rapidly, routinely providing reliable variant calls in most of the human genome. We describe how advances in long reads, deep learning, de novo assembly and pangenomes have expanded access to variant calls in increasingly challenging, repetitive genomic regions, including medically relevant regions, and how new benchmark sets and benchmarking methods illuminate their strengths and limitations. Finally, we explore the possible future of more complete characterization of human genome variation in light of the recent completion of a telomere-to-telomere human genome reference assembly and human pangenomes, and we consider the innovations needed to benchmark their newly accessible repetitive regions and complex variants.

DeepSom: a CNN-based approach to somatic variant calling in WGS samples without a matched normal

MOTIVATION

Somatic mutations are usually called by analyzing the DNA sequence of a tumor sample in conjunction with a matched normal. However, a matched normal is not always available, for instance, in retrospective analysis or diagnostic settings. For such cases, tumor-only somatic variant calling tools need to be designed. Previously proposed approaches demonstrate inferior performance on whole-genome sequencing (WGS) samples.

RESULTS

We present the convolutional neural network-based approach called DeepSom for detecting somatic single nucleotide polymorphism and short insertion and deletion variants in tumor WGS samples without a matched normal. We validate DeepSom by reporting its performance on five different cancer datasets. We also demonstrate that on WGS samples DeepSom outperforms previously proposed methods for tumor-only somatic variant calling.

AVAILABILITY AND IMPLEMENTATION

DeepSom is available as a GitHub repository at https://github.com/heiniglab/DeepSom.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Artificial intelligence applied in neoantigen identification facilitates personalized cancer immunotherapy

The field of cancer neoantigen investigation has developed swiftly in the past decade. Predicting novel and true neoantigens derived from large multi-omics data became difficult but critical challenges. The rise of Artificial Intelligence (AI) or Machine Learning (ML) in biomedicine application has brought benefits to strengthen the current computational pipeline for neoantigen prediction. ML algorithms offer powerful tools to recognize the multidimensional nature of the omics data and therefore extract the key neoantigen features enabling a successful discovery of new neoantigens. The present review aims to outline the significant technology progress of machine learning approaches, especially the newly deep learning tools and pipelines, that were recently applied in neoantigen prediction. In this review article, we summarize the current state-of-the-art tools developed to predict neoantigens. The standard workflow includes calling genetic variants in paired tumor and blood samples, and rating the binding affinity between mutated peptide, MHC (I and II) and T cell receptor (TCR), followed by characterizing the immunogenicity of tumor epitopes. More specifically, we highlight the outstanding feature extraction tools and multi-layer neural network architectures in typical ML models. It is noted that more integrated neoantigen-predicting pipelines are constructed with hybrid or combined ML algorithms instead of conventional machine learning models. In addition, the trends and challenges in further optimizing and integrating the existing pipelines are discussed.

Accurate somatic variant detection using weakly supervised deep learning

Identification of somatic mutations in tumor samples is commonly based on statistical methods in combination with heuristic filters. Here we develop VarNet, an end-to-end deep learning approach for identification of somatic variants from aligned tumor and matched normal DNA reads. VarNet is trained using image representations of 4.6 million high-confidence somatic variants annotated in 356 tumor whole genomes. We benchmark VarNet across a range of publicly available datasets, demonstrating performance often exceeding current state-of-the-art methods. Overall, our results demonstrate how a scalable deep learning approach could augment and potentially supplant human engineered features and heuristic filters in somatic variant calling.

Development and performance evaluation of an artificial intelligence algorithm using cell-free DNA fragment distance for non-invasive prenatal testing (aiD-NIPT)

With advances in next-generation sequencing technology, non-invasive prenatal testing (NIPT) has been widely implemented to detect fetal aneuploidies, including trisomy 21, 18, and 13 (T21, T18, and T13). Most NIPT methods use cell-free DNA (cfDNA) fragment count (FC) in maternal blood. In this study, we developed a novel NIPT method using cfDNA fragment distance (FD) and convolutional neural network-based artificial intelligence algorithm (aiD-NIPT). Four types of aiD-NIPT algorithm (mean, median, interquartile range, and its ensemble) were developed using 2,215 samples. In an analysis of 17,678 clinical samples, all algorithms showed >99.40% accuracy for T21/T18/T13, and the ensemble algorithm showed the best performance (sensitivity: 99.07%, positive predictive value (PPV): 88.43%); the FC-based conventional Z-score and normalized chromosomal value showed 98.15% sensitivity, with 40.77% and 36.81% PPV, respectively. In conclusion, FD-based aiD-NIPT was successfully developed, and it showed better performance than FC-based NIPT methods.

STAG2 expression is associated with adverse survival outcomes and regulates cell phenotype in muscle-invasive bladder cancer

STAG2 (Stromal Antigen 2), in healthy somatic cells, functions in sister chromatid cohesion, DNA damage repair, and genome organization, but its role in muscle invasive bladder cancer (MIBC) remains unknown. Here, using whole-exome and targeted sequencing (n=119 bladder cancer clinical samples), we found several STAG2 mutations in MIBC that correlate with loss of protein expression. The analysis of a bladder cancer tissue microarray (n=346) revealed that decreased STAG2 protein expression is associated with improved overall and progression-free survival for MIBC patients. In mouse xenograft studies, STAG2 knockdown (KD) decelerated MIBC tumor growth, whereas STAG2 overexpression accelerated tumor growth. In cell line studies, STAG2 loss augmented treatment with cisplatin, a first-line therapy for MIBC. STAG2 KD or overexpression did not alter degree of aneuploidy, copy number variations, or cell cycle distribution. However, unbiased RNA sequencing analysis revealed that STAG2 KD altered gene expression. STAG2 KD led to significant downregulation of several gene sets, such as collagen containing extracellular matrix, external encapsulating structure organization, and regulation of chemotaxis. Therefore, we investigated the effect of STAG2 KD on cell migration and invasion in vitro. We found that STAG2 KD minimized cell speed, displacement, and invasion. Altogether, our results present a non-canonical function of STAG2 in promoting cell motility and invasion of MIBC cells. This work forms the basis for additional investigation into the role of STAG2 in transcriptional regulation and how it becomes dysregulated in STAG2-mutant MIBC.

DETexT: An SNV detection enhancement for low read depth by integrating mutational signatures into TextCNN

Detecting SNV at very low read depths helps to reduce sequencing requirements, lowers sequencing costs, and aids in the early screening, diagnosis, and treatment of cancer. However, the accuracy of SNV detection is significantly reduced at read depths below ×34 due to the lack of a sufficient number of read pairs to help filter out false positives. Many recent studies have revealed the potential of mutational signature (MS) in detecting true SNV, understanding the mutational processes that lead to the development of human cancers, and analyzing the endogenous and exogenous causes. Here, we present DETexT, an SNV detection method better suited to low read depths, which classifies false positive variants by combining MS with deep learning algorithms to mine correlation information around bases in individual reads without relying on the support of duplicate read pairs. We have validated the effectiveness of DETexT on simulated and real datasets and conducted comparative experiments. The source code has been uploaded to https://github.com/TrinaZ/extra-lowRD for academic use only.

Analysis of low-level somatic mosaicism reveals stage and tissue-specific mutational features in human development

Most somatic mutations that arise during normal development are present at low levels in single or multiple tissues depending on the developmental stage and affected organs. However, the effect of human developmental stages or mutations of different organs on the features of somatic mutations is still unclear. Here, we performed a systemic and comprehensive analysis of low-level somatic mutations using deep whole-exome sequencing (average read depth ~500×) of 498 multiple organ tissues with matched controls from 190 individuals. Our results showed that early clone-forming mutations shared between multiple organs were lower in number but showed higher allele frequencies than late clone-forming mutations [0.54 vs. 5.83 variants per individual; 6.17% vs. 1.5% variant allele frequency (VAF)] along with less nonsynonymous mutations and lower functional impacts. Additionally, early and late clone-forming mutations had unique mutational signatures that were distinct from mutations that originated from tumors. Compared with early clone-forming mutations that showed a clock-like signature across all organs or tissues studied, late clone-forming mutations showed organ, tissue, and cell-type specificity in the mutation counts, VAFs, and mutational signatures. In particular, analysis of brain somatic mutations showed a bimodal occurrence and temporal-lobe-specific signature. These findings provide new insights into the features of somatic mosaicism that are dependent on developmental stage and brain regions.

Most somatic mutations that arise during normal development are present at low levels in single or multiple tissues, and often show a degree of clonality depending on the time and origin of the mutation. Recent studies have identified the characteristics of postzygotic variants of somatic mutations at the single-cell or mono-clonal levels. However, the results may not be fully representative of the mutational processes involved. Here, we describe a comprehensive analysis of low-level somatic mutations identified after deep whole-exome sequencing in peripheral and brain tissues. We found that clone-forming mutations are uniquely defined by early and late-stage aspects in the mutational profiles. Thus, we identified reliable spatiotemporal characteristics of mosaic variants. Additionally, we found low-level clone-forming mosaic variants across multiple stages and tissues, and identified their intrinsic features.

Bioinformatics: From NGS Data to Biological Complexity in Variant Detection and Oncological Clinical Practice

The use of next-generation sequencing (NGS) techniques for variant detection has become increasingly important in clinical research and in clinical practice in oncology. Many cancer patients are currently being treated in clinical practice or in clinical trials with drugs directed against specific genomic alterations. In this scenario, the development of reliable and reproducible bioinformatics tools is essential to derive information on the molecular characteristics of each patient’s tumor from the NGS data. The development of bioinformatics pipelines based on the use of machine learning and statistical methods is even more relevant for the determination of complex biomarkers. In this review, we describe some important technologies, computational algorithms and models that can be applied to NGS data from Whole Genome to Targeted Sequencing, to address the problem of finding complex cancer-associated biomarkers. In addition, we explore the future perspectives and challenges faced by bioinformatics for precision medicine both at a molecular and clinical level, with a focus on an emerging complex biomarker such as homologous recombination deficiency (HRD).

RDscan: A New Method for Improving Germline and Somatic Variant Calling Based on Read Depth Distribution

Several tools have been developed for calling variants from next-generation sequencing (NGS) data. Although they are generally accurate and reliable, most of them have room for improvement, especially regarding calling variants in datasets with low read depth. In addition, the somatic variants predicted by several somatic variant callers tend to have very low concordance rates. In this study, we developed a new method (RDscan) for improving germline and somatic variant calling in NGS data. RDscan removes misaligned reads, repositions reads, and calculates RDscore based on the read depth distribution. With RDscore, RDscan improves the precision of variant callers by removing false-positive variant calls. When we tested our new tool using the latest variant calling algorithms and data from the 1000 Genomes Project and Illumina's public datasets, accuracy was improved for most of the algorithms. After screening variants with RDscan, calling accuracies increased for germline variants in 11 of 12 cases and for somatic variants in 21 of 24 cases. RDscan is simple to use and can effectively remove false-positive variants while maintaining a low computation load. Therefore, RDscan, along with existing variant callers, should contribute to improvements in genome analysis.

Artificial intelligence and machine learning in precision and genomic medicine

The advancement of precision medicine in medical care has led behind the conventional symptom-driven treatment process by allowing early risk prediction of disease through improved diagnostics and customization of more effective treatments. It is necessary to scrutinize overall patient data alongside broad factors to observe and differentiate between ill and relatively healthy people to take the most appropriate path toward precision medicine, resulting in an improved vision of biological indicators that can signal health changes. Precision and genomic medicine combined with artificial intelligence have the potential to improve patient healthcare. Patients with less common therapeutic responses or unique healthcare demands are using genomic medicine technologies. AI provides insights through advanced computation and inference, enabling the system to reason and learn while enhancing physician decision making. Many cell characteristics, including gene up-regulation, proteins binding to nucleic acids, and splicing, can be measured at high throughput and used as training objectives for predictive models. Researchers can create a new era of effective genomic medicine with the improved availability of a broad range of datasets and modern computer techniques such as machine learning. This review article has elucidated the contributions of ML algorithms in precision and genome medicine.

Artificial intelligence and machine learning in precision and genomic medicine

The advancement of precision medicine in medical care has led behind the conventional symptom-driven treatment process by allowing early risk prediction of disease through improved diagnostics and customization of more effective treatments. It is necessary to scrutinize overall patient data alongside broad factors to observe and differentiate between ill and relatively healthy people to take the most appropriate path toward precision medicine, resulting in an improved vision of biological indicators that can signal health changes. Precision and genomic medicine combined with artificial intelligence have the potential to improve patient healthcare. Patients with less common therapeutic responses or unique healthcare demands are using genomic medicine technologies. AI provides insights through advanced computation and inference, enabling the system to reason and learn while enhancing physician decision making. Many cell characteristics, including gene up-regulation, proteins binding to nucleic acids, and splicing, can be measured at high throughput and used as training objectives for predictive models. Researchers can create a new era of effective genomic medicine with the improved availability of a broad range of datasets and modern computer techniques such as machine learning. This review article has elucidated the contributions of ML algorithms in precision and genome medicine.

Using generative adversarial networks for genome variant calling from low depth ONT sequencing data

Genome variant calling is a challenging yet critical task for subsequent studies. Existing methods almost rely on high depth DNA sequencing data. Performance on low depth data drops a lot. Using public Oxford Nanopore (ONT) data of human being from the Genome in a Bottle (GIAB) Consortium, we trained a generative adversarial network for low depth variant calling. Our method, noted as LDV-Caller, can project high depth sequencing information from low depth data. It achieves 94.25% F1 score on low depth data, while the F1 score of the state-of-the-art method on two times higher depth data is 94.49%. By doing so, the price of genome-wide sequencing examination can reduce deeply. In addition, we validated the trained LDV-Caller model on 157 public Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) samples. The mean sequencing depth of these samples is 2982. The LDV-Caller yields 92.77% F1 score using only 22x sequencing depth, which demonstrates our method has potential to analyze different species with only low depth sequencing data.

DLEB: a web application for building deep learning models in biological research

Deep learning has been applied for solving many biological problems, and it has shown outstanding performance. Applying deep learning in research requires knowledge of deep learning theories and programming skills, but researchers have developed diverse deep learning platforms to allow users to build deep learning models without programming. Despite these efforts, it is still difficult for biologists to use deep learning because of limitations of the existing platforms. Therefore, a new platform is necessary that can solve these challenges for biologists. To alleviate this situation, we developed a user-friendly and easy-to-use web application called DLEB (Deep Learning Editor for Biologists) that allows for building deep learning models specialized for biologists. DLEB helps researchers (i) design deep learning models easily and (ii) generate corresponding Python code to run directly in their machines. DLEB provides other useful features for biologists, such as recommending deep learning models for specific learning tasks and data, pre-processing of input biological data, and availability of various template models and example biological datasets for model training. DLEB can serve as a highly valuable platform for easily applying deep learning to solve many important biological problems. DLEB is freely available at http://dleb.konkuk.ac.kr/.

DLSSAffinity: protein-ligand binding affinity prediction via a deep learning model

Evaluating the protein-ligand binding affinity is a substantial part of the computer-aided drug discovery process. Most of the proposed computational methods predict protein-ligand binding affinity using either limited full-length protein 3D structures or simple full-length protein sequences as the input features. Thus, protein-ligand binding affinity prediction remains a fundamental challenge in drug discovery. In this study, we proposed a novel deep learning-based approach, DLSSAffinity, to accurately predict the protein-ligand binding affinity. Unlike the existing methods, DLSSAffinity uses the pocket-ligand structural pairs as the local information to predict short-range direct interactions. Besides, DLSSAffinity also uses the full-length protein sequence and ligand SMILES as the global information to predict long-range indirect interactions. We tested DLSSAffinity on the PDBbind benchmark. The results showed that DLSSAffinity achieves Pearson's R = 0.79, RMSE = 1.40, and SD = 1.35 on the test set. Comparing DLSSAffinity with the existing state-of-the-art deep learning-based binding affinity prediction methods, the DLSSAffinity model outperforms other models. These results demonstrate that combining global sequence and local structure information as the input features of a deep learning model can improve the accuracy of protein-ligand binding affinity prediction.

Curated variation benchmarks for challenging medically relevant autosomal genes

The repetitive nature and complexity of some medically relevant genes poses a challenge for their accurate analysis in a clinical setting. The Genome in a Bottle Consortium has provided variant benchmark sets, but these exclude nearly 400 medically relevant genes due to their repetitiveness or polymorphic complexity. Here, we characterize 273 of these 395 challenging autosomal genes using a haplotype-resolved whole-genome assembly. This curated benchmark reports over 17,000 single-nucleotide variations, 3,600 insertions and deletions and 200 structural variations each for human genome reference GRCh37 and GRCh38 across HG002. We show that false duplications in either GRCh37 or GRCh38 result in reference-specific, missed variants for short- and long-read technologies in medically relevant genes, including CBS, CRYAA and KCNE1. When masking these false duplications, variant recall can improve from 8% to 100%. Forming benchmarks from a haplotype-resolved whole-genome assembly may become a prototype for future benchmarks covering the whole genome.

Computational analysis of cancer genome sequencing data

Distilling biologically meaningful information from cancer genome sequencing data requires comprehensive identification of somatic alterations using rigorous computational methods. As the amount and complexity of sequencing data have increased, so has the number of tools for analysing them. Here, we describe the main steps involved in the bioinformatic analysis of cancer genomes, review key algorithmic developments and highlight popular tools and emerging technologies. These tools include those that identify point mutations, copy number alterations, structural variations and mutational signatures in cancer genomes. We also discuss issues in experimental design, the strengths and limitations of sequencing modalities and methodological challenges for the future.

Identification of neoantigens for individualized therapeutic cancer vaccines

Somatic mutations in cancer cells can generate tumour-specific neoepitopes, which are recognized by autologous T cells in the host. As neoepitopes are not subject to central immune tolerance and are not expressed in healthy tissues, they are attractive targets for therapeutic cancer vaccines. Because the vast majority of cancer mutations are unique to the individual patient, harnessing the full potential of this rich source of targets requires individualized treatment approaches. Many computational algorithms and machine-learning tools have been developed to identify mutations in sequence data, to prioritize those that are more likely to be recognized by T cells and to design tailored vaccines for every patient. In this Review, we fill the gaps between the understanding of basic mechanisms of T cell recognition of neoantigens and the computational approaches for discovery of somatic mutations and neoantigen prediction for cancer immunotherapy. We present a new classification of neoantigens, distinguishing between guarding, restrained and ignored neoantigens, based on how they confer proficient antitumour immunity in a given clinical context. Such context-based differentiation will contribute to a framework that connects neoantigen biology to the clinical setting and medical peculiarities of cancer, and will enable future neoantigen-based therapies to provide greater clinical benefit.

Achieving robust somatic mutation detection with deep learning models derived from reference data sets of a cancer sample

BACKGROUND

Accurate detection of somatic mutations is challenging but critical in understanding cancer formation, progression, and treatment. We recently proposed NeuSomatic, the first deep convolutional neural network-based somatic mutation detection approach, and demonstrated performance advantages on in silico data.

RESULTS

In this study, we use the first comprehensive and well-characterized somatic reference data sets from the SEQC2 consortium to investigate best practices for using a deep learning framework in cancer mutation detection. Using the high-confidence somatic mutations established for a cancer cell line by the consortium, we identify the best strategy for building robust models on multiple data sets derived from samples representing real scenarios, for example, a model trained on a combination of real and spike-in mutations had the highest average performance.

CONCLUSIONS

The strategy identified in our study achieved high robustness across multiple sequencing technologies for fresh and FFPE DNA input, varying tumor/normal purities, and different coverages, with significant superiority over conventional detection approaches in general, as well as in challenging situations such as low coverage, low variant allele frequency, DNA damage, and difficult genomic regions.

CLINICOPATHOLOGIC AND MOLECULAR ANALYSIS OF A SINGLE BCOR-CCNB3+ UNDIFFERENTIATED SARCOMA OF THE KIDNEY CONFERS SIGNIFICANT EPIGENETIC ALTERATIONS

Undifferentiated soft tissue sarcomas (UDSTSs) are a group of mesenchymal tumors that remain a diagnostic challenge because of their morphologic heterogeneity and unclear histologic origin (Peters et al., Mod Pathol28: 575 [2015]). In this case report, we present the first multiomics molecular signature for a BCOR–CCNB3 sarcoma (BCS) that includes mutation analysis, gene expression, DNA methylation, and micro RNA (miRNA) expression. We identify a paucity of additional mutations in this tumor and detail that there is significant dysregulation of gene expression of epigenetic remodeling agents including key members of the PRC, Sin3A/3b, NuRD, and NcoR/SMRT complexes and the DNA methyltransferases DNMT1, DNMT3a, and DNMT3b. This is accompanied by significant DNA methylation changes and dysregulation of multiple miRNAs with known links to tumorigenesis. This study significantly increases our understanding of the BCOR effects on fusion-positive undifferentiated sarcomas at both the genomic and epigenomic level and suggests that as better-tailored and more refined treatment algorithms continue to evolve, epigenetic modifying agents should be further evaluated for their efficacy against these tumors.

Artificial Intelligence as Accelerator for Genomic Medicine and Planetary Health

Genomic medicine has made important strides over the past several decades, but as new insights and technologies emerge, the applications of genomics in medicine and planetary health continue to evolve and expand. An important grand challenge is harnessing and making sense of the genomic big data in ways that best serve public and planetary health. Because human health is inextricably intertwined with the health of planetary ecosystems and nonhuman animals, genomic medicine is in need of high throughput bioinformatics analyses to harness and integrate human and ecological multiomics big data. It is in this overarching context that artificial intelligence (AI), particularly machine learning and deep learning, offers enormous potentials to advance genomic medicine in a spirit of One Health. This expert review offers an analysis of the rapidly emerging role of AI in genomic medicine, including its current drivers, levers, opportunities, and challenges. The scope of AI applications in genomic medicine is broad, ranging from efficient and automated data analysis to drug repurposing and precision medicine, as with its challenges such as veracity of the big data that AI sorely depends on, social biases that the AI-driven algorithms can introduce, and how best to incorporate AI with human intelligence. The road ahead for AI in genomic medicine is complex and arduous and yet worthy of cautious optimism as we face future pandemics and ecological crises in the 21st century. Now is a good time to think about the role of AI in genomic medicine and planetary health.

Calling small variants using universality with Bayes-factor-adjusted odds ratios

The application of next-generation sequencing in research and particularly in clinical routine requires highly accurate variant calling. Here we describe UVC, a method for calling small variants of germline or somatic origin. By unifying opposite assumptions with sublation, we discovered the following two empirical laws to improve variant calling: allele fraction at high sequencing depth is inversely proportional to the cubic root of variant-calling error rate, and odds ratios adjusted with Bayes factors can model various sequencing biases. UVC outperformed other variant callers on the GIAB germline truth sets, 192 scenarios of in silico mixtures simulating 192 combinations of tumor/normal sequencing depths and tumor/normal purities, the GIAB somatic truth sets derived from physical mixture, and the SEQC2 somatic reference sets derived from the breast-cancer cell-line HCC1395. UVC achieved 100% concordance with the manual review conducted by multiple independent researchers on a Qiagen 71-gene-panel dataset derived from 16 patients with colon adenoma. UVC outperformed other unique molecular identifier (UMI)-aware variant callers on the datasets used for publishing these variant callers. Performance was measured with sensitivity-specificity trade off for called variants. The improved variant calls generated by UVC from previously published UMI-based sequencing data provided additional insight about DNA damage repair. UVC is open-sourced under the BSD 3-Clause license at https://github.com/genetronhealth/uvc and quay.io/genetronhealth/gcc-6-3-0-uvc-0-6-0-441a694.

STIC: Predicting Single Nucleotide Variants and Tumor Purity in Cancer Genome

Single nucleotide variant (SNV) plays an important role in cellular proliferation and tumorigenesis in various types of human cancer. Next-generation sequencing (NGS) has provided high-throughput data at an unprecedented resolution to predict SNVs. Currently, there exist many computational methods for either germline or somatic SNV discovery from NGS data, but very few of them are versatile enough to adapt to any situations. In the absence of matched normal samples, the prediction of somatic SNVs from single-tumor samples becomes considerably challenging, especially when the tumor purity is unknown. Here, we propose a new approach, STIC, to predict somatic SNVs and estimate tumor purity from NGS data without matched normal samples. The main features of STIC include: (1) extracting a set of SNV-relevant features on each site and training the BP neural network algorithm on the features to predict SNVs; (2) creating an iterative process to distinguish somatic SNVs from germline ones by disturbing allele frequency; and (3) establishing a reasonable relationship between tumor purity and allele frequencies of somatic SNVs to accurately estimate the purity. We quantitatively evaluate the performance of STIC on both simulation and real sequencing datasets, the results of which indicate that STIC outperforms competing methods.

VC@Scale: Scalable and high-performance variant calling on cluster environments

BACKGROUND

Recently many new deep learning-based variant-calling methods like DeepVariant have emerged as more accurate compared with conventional variant-calling algorithms such as GATK HaplotypeCaller, Sterlka2, and Freebayes albeit at higher computational costs. Therefore, there is a need for more scalable and higher performance workflows of these deep learning methods. Almost all existing cluster-scaled variant-calling workflows that use Apache Spark/Hadoop as big data frameworks loosely integrate existing single-node pre-processing and variant-calling applications. Using Apache Spark just for distributing/scheduling data among loosely coupled applications or using I/O-based storage for storing the output of intermediate applications does not exploit the full benefit of Apache Spark in-memory processing. To achieve this, we propose a native Spark-based workflow that uses Python and Apache Arrow to enable efficient transfer of data between different workflow stages. This benefits from the ease of programmability of Python and the high efficiency of Arrow's columnar in-memory data transformations.

RESULTS

Here we present a scalable, parallel, and efficient implementation of next-generation sequencing data pre-processing and variant-calling workflows. Our design tightly integrates most pre-processing workflow stages, using Spark built-in functions to sort reads by coordinates and mark duplicates efficiently. Our approach outperforms state-of-the-art implementations by >2 times for the pre-processing stages, creating a scalable and high-performance solution for DeepVariant for both CPU-only and CPU + GPU clusters.

CONCLUSIONS

We show the feasibility and easy scalability of our approach to achieve high performance and efficient resource utilization for variant-calling analysis on high-performance computing clusters using the standardized Apache Arrow data representations. All codes, scripts, and configurations used to run our implementations are publicly available and open sourced; see https://github.com/abs-tudelft/variant-calling-at-scale.

Establishing community reference samples, data and call sets for benchmarking cancer mutation detection using whole-genome sequencing

The lack of samples for generating standardized DNA datasets for setting up a sequencing pipeline or benchmarking the performance of different algorithms limits the implementation and uptake of cancer genomics. Here, we describe reference call sets obtained from paired tumor-normal genomic DNA (gDNA) samples derived from a breast cancer cell line-which is highly heterogeneous, with an aneuploid genome, and enriched in somatic alterations-and a matched lymphoblastoid cell line. We partially validated both somatic mutations and germline variants in these call sets via whole-exome sequencing (WES) with different sequencing platforms and targeted sequencing with >2,000-fold coverage, spanning 82% of genomic regions with high confidence. Although the gDNA reference samples are not representative of primary cancer cells from a clinical sample, when setting up a sequencing pipeline, they not only minimize potential biases from technologies, assays and informatics but also provide a unique resource for benchmarking 'tumor-only' or 'matched tumor-normal' analyses.

Convolutional neural networks (CNNs): concepts and applications in pharmacogenomics

Convolutional neural networks (CNNs) have been used to extract information from various datasets of different dimensions. This approach has led to accurate interpretations in several subfields of biological research, like pharmacogenomics, addressing issues previously faced by other computational methods. With the rising attention for personalized and precision medicine, scientists and clinicians have now turned to artificial intelligence systems to provide them with solutions for therapeutics development. CNNs have already provided valuable insights into biological data transformation. Due to the rise of interest in precision and personalized medicine, in this review, we have provided a brief overview of the possibilities of implementing CNNs as an effective tool for analyzing one-dimensional biological data, such as nucleotide and protein sequences, as well as small molecular data, e.g., simplified molecular-input line-entry specification, InChI, binary fingerprints, etc., to categorize the models based on their objective and also highlight various challenges. The review is organized into specific research domains that participate in pharmacogenomics for a more comprehensive understanding. Furthermore, the future intentions of deep learning are outlined.

Integrated Analysis of Whole Genome and Epigenome Data Using Machine Learning Technology: Toward the Establishment of Precision Oncology

With the completion of the International Human Genome Project, we have entered what is known as the post-genome era, and efforts to apply genomic information to medicine have become more active. In particular, with the announcement of the Precision Medicine Initiative by U.S. President Barack Obama in his State of the Union address at the beginning of 2015, "precision medicine," which aims to divide patients and potential patients into subgroups with respect to disease susceptibility, has become the focus of worldwide attention. The field of oncology is also actively adopting the precision oncology approach, which is based on molecular profiling, such as genomic information, to select the appropriate treatment. However, the current precision oncology is dominated by a method called targeted-gene panel (TGP), which uses next-generation sequencing (NGS) to analyze a limited number of specific cancer-related genes and suggest optimal treatments, but this method causes the problem that the number of patients who benefit from it is limited. In order to steadily develop precision oncology, it is necessary to integrate and analyze more detailed omics data, such as whole genome data and epigenome data. On the other hand, with the advancement of analysis technologies such as NGS, the amount of data obtained by omics analysis has become enormous, and artificial intelligence (AI) technologies, mainly machine learning (ML) technologies, are being actively used to make more efficient and accurate predictions. In this review, we will focus on whole genome sequencing (WGS) analysis and epigenome analysis, introduce the latest results of omics analysis using ML technologies for the development of precision oncology, and discuss the future prospects.

From rumor to genetic mutation detection with explanations: a GAN approach

Social media have emerged as increasingly popular means and environments for information gathering and propagation. This vigorous growth of social media contributed not only to a pandemic (fast-spreading and far-reaching) of rumors and misinformation, but also to an urgent need for text-based rumor detection strategies. To speed up the detection of misinformation, traditional rumor detection methods based on hand-crafted feature selection need to be replaced by automatic artificial intelligence (AI) approaches. AI decision making systems require to provide explanations in order to assure users of their trustworthiness. Inspired by the thriving development of generative adversarial networks (GANs) on text applications, we propose a GAN-based layered model for rumor detection with explanations. To demonstrate the universality of the proposed approach, we demonstrate its benefits on a gene classification with mutation detection case study. Similarly to the rumor detection, the gene classification can also be formulated as a text-based classification problem. Unlike fake news detection that needs a previously collected verified news database, our model provides explanations in rumor detection based on tweet-level texts only without referring to a verified news database. The layered structure of both generative and discriminative models contributes to the outstanding performance. The layered generators produce rumors by intelligently inserting controversial information in non-rumors, and force the layered discriminators to detect detailed glitches and deduce exactly which parts in the sentence are problematic. On average, in the rumor detection task, our proposed model outperforms state-of-the-art baselines on PHEME dataset by \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$26.85\%$$\end{document}26.85% in terms of macro-f1. The excellent performance of our model for textural sequences is also demonstrated by the gene mutation case study on which it achieves \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$72.69\%$$\end{document}72.69% macro-f1 score.

Deep learning in next-generation sequencing

Next-generation sequencing (NGS) methods lie at the heart of large parts of biological and medical research. Their fundamental importance has created a continuously increasing demand for processing and analysis methods of the data sets produced, addressing questions such as variant calling, metagenomic classification and quantification, genomic feature detection, or downstream analysis in larger biological or medical contexts. In addition to classical algorithmic approaches, machine-learning (ML) techniques are often used for such tasks. In particular, deep learning (DL) methods that use multilayered artificial neural networks (ANNs) for supervised, semisupervised, and unsupervised learning have gained significant traction for such applications. Here, we highlight important network architectures, application areas, and DL frameworks in a NGS context.

Semi-supervised learning for somatic variant calling and peptide identification in personalized cancer immunotherapy

BACKGROUND

Personalized cancer vaccines are emerging as one of the most promising approaches to immunotherapy of advanced cancers. However, only a small proportion of the neoepitopes generated by somatic DNA mutations in cancer cells lead to tumor rejection. Since it is impractical to experimentally assess all candidate neoepitopes prior to vaccination, developing accurate methods for predicting tumor-rejection mediating neoepitopes (TRMNs) is critical for enabling routine clinical use of cancer vaccines.

RESULTS

In this paper we introduce Positive-unlabeled Learning using AuTOml (PLATO), a general semi-supervised approach to improving accuracy of model-based classifiers. PLATO generates a set of high confidence positive calls by applying a stringent filter to model-based predictions, then rescores remaining candidates by using positive-unlabeled learning. To achieve robust performance on clinical samples with large patient-to-patient variation, PLATO further integrates AutoML hyper-parameter tuning, classification threshold selection based on spies, and support for bootstrapping.

CONCLUSIONS

Experimental results on real datasets demonstrate that PLATO has improved performance compared to model-based approaches for two key steps in TRMN prediction, namely somatic variant calling from exome sequencing data and peptide identification from MS/MS data.

Training confounder-free deep learning models for medical applications

The presence of confounding effects (or biases) is one of the most critical challenges in using deep learning to advance discovery in medical imaging studies. Confounders affect the relationship between input data (e.g., brain MRIs) and output variables (e.g., diagnosis). Improper modeling of those relationships often results in spurious and biased associations. Traditional machine learning and statistical models minimize the impact of confounders by, for example, matching data sets, stratifying data, or residualizing imaging measurements. Alternative strategies are needed for state-of-the-art deep learning models that use end-to-end training to automatically extract informative features from large set of images. In this article, we introduce an end-to-end approach for deriving features invariant to confounding factors while accounting for intrinsic correlations between the confounder(s) and prediction outcome. The method does so by exploiting concepts from traditional statistical methods and recent fair machine learning schemes. We evaluate the method on predicting the diagnosis of HIV solely from Magnetic Resonance Images (MRIs), identifying morphological sex differences in adolescence from those of the National Consortium on Alcohol and Neurodevelopment in Adolescence (NCANDA), and determining the bone age from X-ray images of children. The results show that our method can accurately predict while reducing biases associated with confounders. The code is available at https://github.com/qingyuzhao/br-net .