Extend the benchmarking indel set by manual review using the individual cell line sequencing data from the Sequencing Quality Control 2 (SEQC2) project

Accurate indel calling plays an important role in precision medicine. A benchmarking indel set is essential for thoroughly evaluating the indel calling performance of bioinformatics pipelines. A reference sample with a set of known-positive variants was developed in the FDA-led Sequencing Quality Control Phase 2 (SEQC2) project, but the known indels in the known-positive set were limited. This project sought to provide an enriched set of known indels that would be more translationally relevant by focusing on additional cancer related regions. A thorough manual review process completed by 42 reviewers, two advisors, and a judging panel of three researchers significantly enriched the known indel set by an additional 516 indels. The extended benchmarking indel set has a large range of variant allele frequencies (VAFs), with 87% of them having a VAF below 20% in reference Sample A. The reference Sample A and the indel set can be used for comprehensive benchmarking of indel calling across a wider range of VAF values in the lower range. Indel length was also variable, but the majority were under 10 base pairs (bps). Most of the indels were within coding regions, with the remainder in the gene regulatory regions. Although high confidence can be derived from the robust study design and meticulous human review, this extensive indel set has not undergone orthogonal validation. The extended benchmarking indel set, along with the indels in the previously published known-positive set, was the truth set used to benchmark indel calling pipelines in a community challenge hosted on the precisionFDA platform. This benchmarking indel set and reference samples can be utilized for a comprehensive evaluation of indel calling pipelines. Additionally, the insights and solutions obtained during the manual review process can aid in improving the performance of these pipelines.

PredinID: Predicting Pathogenic Inframe Indels in Human Through Graph Convolution Neural Network With Graph Sampling Technique

Inframe insertion/deletion (indel) variants may alter protein sequence and function, which are closely related to an extensive variety of diseases. Although recent researches have paid attention to the associations between inframe indels and diseases, modeling indels in silico and interpreting their pathogenicity remain challenging, mainly due to the lack of experimental information and computational methodologies. In this article, we propose a novel computational method named PredinID (Predictor for inframe InDels) via graph convolutional network (GCN). PredinID leverages k-nearest neighbor algorithm to construct the feature graph for aggregating more informative representation, regarding the pathogenic inframe indel prediction as a node classification task. An edge-based sampling strategy is designed for extracting information from both the potential connections of feature space and the topological structure of subgraphs. Evaluated by 5-fold cross-validations, the PredinID method achieves satisfactory performance and is superior to four classic machine learning algorithms and two GCN methods. Comprehensive experiments show that PredinID has superior performances when compared with the state-of-the-art methods on the independent test set. Moreover, we also implement a web server at http://predinid.bio.aielab.cc/, to facilitate the use of the model.

Structural and functional analysis of somatic coding and UTR indels in breast and lung cancer genomes

Insertions and deletions (Indels) represent one of the major variation types in the human genome and have been implicated in diseases including cancer. To study the features of somatic indels in different cancer genomes, we investigated the indels from two large samples of cancer types: invasive breast carcinoma (BRCA) and lung adenocarcinoma (LUAD). Besides mapping somatic indels in both coding and untranslated regions (UTRs) from the cancer whole exome sequences, we investigated the overlap between these indels and transcription factor binding sites (TFBSs), the key elements for regulation of gene expression that have been found in both coding and non-coding sequences. Compared to the germline indels in healthy genomes, somatic indels contain more coding indels with higher than expected frame-shift (FS) indels in cancer genomes. LUAD has a higher ratio of deletions and higher coding and FS indel rates than BRCA. More importantly, these somatic indels in cancer genomes tend to locate in sequences with important functions, which can affect the core secondary structures of proteins and have a bigger overlap with predicted TFBSs in coding regions than the germline indels. The somatic CDS indels are also enriched in highly conserved nucleotides when compared with germline CDS indels.

DLFF-ACP: prediction of ACPs based on deep learning and multi-view features fusion

An emerging type of therapeutic agent, anticancer peptides (ACPs), has attracted attention because of its lower risk of toxic side effects. However process of identifying ACPs using experimental methods is both time-consuming and laborious. In this study, we developed a new and efficient algorithm that predicts ACPs by fusing multi-view features based on dual-channel deep neural network ensemble model. In the model, one channel used the convolutional neural network CNN to automatically extract the potential spatial features of a sequence. Another channel was used to process and extract more effective features from handcrafted features. Additionally, an effective feature fusion method was explored for the mutual fusion of different features. Finally, we adopted the neural network to predict ACPs based on the fusion features. The performance comparisons across the single and fusion features showed that the fusion of multi-view features could effectively improve the model's predictive ability. Among these, the fusion of the features extracted by the CNN and composition of k-spaced amino acid group pairs achieved the best performance. To further validate the performance of our model, we compared it with other existing methods using two independent test sets. The results showed that our model's area under curve was 0.90, which was higher than that of the other existing methods on the first test set and higher than most of the other existing methods on the second test set. The source code and datasets are available at https://github.com/wame-ng/DLFF-ACP.

dbMCS: A Database for Exploring the Mutation Markers of Anti-Cancer Drug Sensitivity

The identification of mutation markers and the selection of appropriate treatment for patients with specific genome mutations are important steps in the development of targeted therapies and the realization of precision medicine for human cancers. To investigate the baseline characteristics of drug sensitivity markers and develop computational methods of mutation effect prediction, we presented a manually curated online- based database of mutation Markers for anti-Cancer drug Sensitivity (dbMCS). Currently, dbMCS contains 1271 mutations and 4427 mutation-disease-drug associations (3151 and 1276 for sensitivity and resistance, respectively) with their PubMed indexed articles. By comparing the mutations in dbMCS with the putative neutral polymorphisms, we investigated the characteristics of drug sensitivity markers. We found that the mutation markers tend to significantly impact on high-conservative regions both in DNA sequences and protein domains. And some of them presented pleiotropic effects depending on the tumor context, appearing concurrently in the sensitivity and resistance categories. In addition, we preliminarily explored the machine learning-based methods for identifying mutation markers of anti-cancer drug sensitivity and produced optimistic results, which suggests that a reliable dataset may provide new insights and essential clues for future cancer pharmacogenomics studies. dbMCS is available at http://bioinfo.aielab.cc/dbMCS/.

DriverSubNet: A Novel Algorithm for Identifying Cancer Driver Genes by Subnetwork Enrichment Analysis

Identification of driver genes from mass non-functional passenger genes in cancers is still a critical challenge. Here, an effective and no parameter algorithm, named DriverSubNet, is presented for detecting driver genes by effectively mining the mutation and gene expression information based on subnetwork enrichment analysis. Compared with the existing classic methods, DriverSubNet can rank driver genes and filter out passenger genes more efficiently in terms of precision, recall, and F1 score, as indicated by the analysis of four cancer datasets. The method recovered about 50% more known cancer driver genes in the top 100 detected genes than those found in other algorithms. Intriguingly, DriverSubNet was able to find these unknown cancer driver genes which could act as potential therapeutic targets and useful prognostic biomarkers for cancer patients. Therefore, DriverSubNet may act as a useful tool for the identification of driver genes by subnetwork enrichment analysis.

Utility of a custom designed next generation DNA sequencing gene panel to molecularly classify endometrial cancers according to The Cancer Genome Atlas subgroups

BACKGROUND

The Cancer Genome Atlas identified four molecular subgroups of endometrial cancer with survival differences based on whole genome, transcriptomic, and proteomic characterization. Clinically accessible algorithms that reproduce this data are needed. Our aim was to determine if targeted sequencing alone allowed for molecular classification of endometrial cancer.

METHODS

Using a custom-designed 156 gene panel, we analyzed 47 endometrial cancers and matching non-tumor tissue. Variants were annotated for pathogenicity and medical records were reviewed for the clinicopathologic variables. Using molecular characteristics, tumors were classified into four subgroups. Group 1 included patients with > 570 unfiltered somatic variants, > 9 cytosine to adenine nucleotide substitutions per sample, and < 1 cytosine to guanine nucleotide substitution per sample. Group 2 included patients with any somatic mutation in MSH2, MSH6, MLH1, PMS2. Group 3 included patients with TP53 mutations without mutation in mismatch repair genes. Remaining patients were classified as group 4. Analyses were performed using SAS 9.4 (SAS Institute Inc., Cary, North Carolina, USA).

RESULTS

Endometrioid endometrial cancers had more candidate variants of potential pathogenic interest (median 6 IQR 4.13 vs. 2 IQR 2.3; p < 0.01) than uterine serous cancers. PTEN (82% vs. 15%, p < 0.01) and PIK3CA (74% vs. 23%, p < 0.01) mutations were more frequent in endometrioid than serous carcinomas. TP53 (18% vs. 77%, p < 0.01) mutations were more frequent in serous carcinomas. Visual inspection of the number of unfiltered somatic variants per sample identified six grade 3 endometrioid samples with high tumor mutational burden, all of which demonstrated POLE mutations, most commonly P286R and V411L. Of the grade 3 endometrioid carcinomas, those with POLE mutations were less likely to have risk factors necessitating adjuvant treatment than those with low tumor mutational burden. Targeted sequencing was unable to assign samples to microsatellite unstable, copy number low, and copy number high subgroups.

CONCLUSIONS

Targeted sequencing can predict the presence of POLE mutations based on the tumor mutational burden. However, targeted sequencing alone is inadequate to classify endometrial cancers into molecular subgroups identified by The Cancer Genome Atlas.

Predicting Hot Spot Residues at Protein-DNA Binding Interfaces Based on Sequence Information

Hot spot residues at protein-DNA binding interfaces are hugely important for investigating the underlying mechanism of molecular recognition. Currently, there are a few tools available for identifying the hot spot residues in the protein-DNA complexes. In addition, the three-dimensional protein structures are needed in these tools. However, it is well known that the three-dimensional structures are unavailable for most proteins. Considering the limitation, we proposed a method, named SPDH, for predicting hot spot residues only based on protein sequences. Firstly, we obtained 133 features from physicochemical property, conservation, predicted solvent accessible surface area and structure. Then, we systematically assessed these features based on various feature selection methods to obtain the optimal feature subset and compared the models using four classical machine learning algorithms (support vector machine, random forest, logistic regression, and k-nearest neighbor) on the training dataset. We found that the variability of physicochemical property features between wild and mutative types was important on improving the performance of the prediction model. On the independent test set, our method achieved the performance with AUC of 0.760 and sensitivity of 0.808, and outperformed other methods. The data and source code can be downloaded at https://github.com/xialab-ahu/SPDH .

PredCID: prediction of driver frameshift indels in human cancer

The discrimination of driver from passenger mutations has been a hot topic in the field of cancer biology. Although recent advances have improved the identification of driver mutations in cancer genomic research, there is no computational method specific for the cancer frameshift indels (insertions or/and deletions) yet. In addition, existing pathogenic frameshift indel predictors may suffer from plenty of missing values because of different choices of transcripts during the variant annotation processes. In this study, we proposed a computational model, called PredCID (Predictor for Cancer driver frameshift InDels), for accurately predicting cancer driver frameshift indels. Gene, DNA, transcript and protein level features are combined together and selected for classification with eXtreme Gradient Boosting classifier. Benchmarking results on the cross-validation dataset and independent dataset showed that PredCID achieves better and robust performance compared with existing noncancer-specific methods in distinguishing cancer driver frameshift indels from passengers and is therefore a valuable method for deeper understanding of frameshift indels in human cancer. PredCID is freely available for academic research at http://bioinfo.ahu.edu.cn:8080/PredCID.

Pathogenicity and functional impact of non-frameshifting insertion/deletion variation in the human genome

Differentiation between phenotypically neutral and disease-causing genetic variation remains an open and relevant problem. Among different types of variation, non-frameshifting insertions and deletions (indels) represent an understudied group with widespread phenotypic consequences. To address this challenge, we present a machine learning method, MutPred-Indel, that predicts pathogenicity and identifies types of functional residues impacted by non-frameshifting insertion/deletion variation. The model shows good predictive performance as well as the ability to identify impacted structural and functional residues including secondary structure, intrinsic disorder, metal and macromolecular binding, post-translational modifications, allosteric sites, and catalytic residues. We identify structural and functional mechanisms impacted preferentially by germline variation from the Human Gene Mutation Database, recurrent somatic variation from COSMIC in the context of different cancers, as well as de novo variants from families with autism spectrum disorder. Further, the distributions of pathogenicity prediction scores generated by MutPred-Indel are shown to differentiate highly recurrent from non-recurrent somatic variation. Collectively, we present a framework to facilitate the interrogation of both pathogenicity and the functional effects of non-frameshifting insertion/deletion variants. The MutPred-Indel webserver is available at http://mutpred.mutdb.org/.