iFish: predicting the pathogenicity of human nonsynonymous variants using gene-specific/family-specific attributes and classifiers

Comparative analysis of in-silico tools in identifying pathogenic variants in dominant inherited retinal diseases

Inherited retinal diseases (IRDs) are a group of rare genetic eye conditions that cause blindness. Despite progress in identifying genes associated with IRDs, improvements are necessary for classifying rare autosomal dominant (AD) disorders. AD diseases are highly heterogenous, with causal variants being restricted to specific amino acid changes within certain protein domains, making AD conditions difficult to classify. Here, we aim to determine the top-performing in-silico tools for predicting the pathogenicity of AD IRD variants. We annotated variants from ClinVar and benchmarked 39 variant classifier tools on IRD genes, split by inheritance pattern. Using area-under-the-curve (AUC) analysis, we determined the top-performing tools and defined thresholds for variant pathogenicity. Top-performing tools were assessed using genome sequencing on a cohort of participants with IRDs of unknown etiology. MutScore achieved the highest accuracy within AD genes, yielding an AUC of 0.969. When filtering for AD gain-of-function and dominant negative variants, BayesDel had the highest accuracy with an AUC of 0.997. Five participants with variants in NR2E3, RHO, GUCA1A, and GUCY2D were confirmed to have dominantly inherited disease based on pedigree, phenotype, and segregation analysis. We identified two uncharacterized variants in GUCA1A (c.428T>A, p.Ile143Thr) and RHO (c.631C>G, p.His211Asp) in three participants. Our findings support using a multi-classifier approach comprised of new missense classifier tools to identify pathogenic variants in participants with AD IRDs. Our results provide a foundation for improved genetic diagnosis for people with IRDs.

Enhancing missense variant pathogenicity prediction with protein language models using VariPred

Computational approaches for predicting the pathogenicity of genetic variants have advanced in recent years. These methods enable researchers to determine the possible clinical impact of rare and novel variants. Historically these prediction methods used hand-crafted features based on structural, evolutionary, or physiochemical properties of the variant. In this study we propose a novel framework that leverages the power of pre-trained protein language models to predict variant pathogenicity. We show that our approach VariPred (Variant impact Predictor) outperforms current state-of-the-art methods by using an end-to-end model that only requires the protein sequence as input. Using one of the best-performing protein language models (ESM-1b), we establish a robust classifier that requires no calculation of structural features or multiple sequence alignments. We compare the performance of VariPred with other representative models including 3Cnet, Polyphen-2, REVEL, MetaLR, FATHMM and ESM variant. VariPred performs as well as, or in most cases better than these other predictors using six variant impact prediction benchmarks despite requiring only sequence data and no pre-processing of the data.

Predicting deleterious missense genetic variants via integrative supervised nonnegative matrix tri-factorization

Among an assortment of genetic variations, Missense are major ones which a small subset of them may led to the upset of the protein function and ultimately end in human diseases. Various machine learning methods were declared to differentiate deleterious and benign missense variants by means of a large number of features, including structure, sequence, interaction networks, gene disease associations as well as phenotypes. However, development of a reliable and accurate algorithm for merging heterogeneous information is highly needed as it could be captured all information of complex interactions on network that genes participate in. In this study we proposed a new method based on the non-negative matrix tri-factorization clustering method. We outlined two versions of the proposed method: two-source and three-source algorithms. Two-source algorithm aggregates individual deleteriousness prediction methods and PPI network, and three-source algorithm incorporates gene disease associations into the other sources already mentioned. Four benchmark datasets were employed for internally and externally validation of both algorithms of our predictor. The results at all datasets confirmed that, our method outperforms most state of the art variant prediction tools. Two key features of our variant effect prediction method are worth mentioning. Firstly, despite the fact that the incorporation of gene disease information at three-source algorithm can improve prediction performance by comparison with two-source algorithm, our method did not hinder by type 2 circularity error unlike some recent ensemble-based prediction methods. Type 2 circularity error occurs when the predictor annotates variants on the basis of the genes located on. Secondly, the performance of our predictor is superior over other ensemble-based methods for variants positioned on genes in which we do not have enough information about their pathogenicity.

Developmental and temporal characteristics of clonal sperm mosaicism

Throughout development and aging, human cells accumulate mutations resulting in genomic mosaicism and genetic diversity at the cellular level. Mosaic mutations present in the gonads can affect both the individual and the offspring and subsequent generations. Here, we explore patterns and temporal stability of clonal mosaic mutations in male gonads by sequencing ejaculated sperm. Through 300× whole-genome sequencing of blood and sperm from healthy men, we find each ejaculate carries on average 33.3 ± 12.1 (mean ± SD) clonal mosaic variants, nearly all of which are detected in serial sampling, with the majority absent from sampled somal tissues. Their temporal stability and mutational signature suggest origins during embryonic development from a largely immutable stem cell niche. Clonal mosaicism likely contributes a transmissible, predicted pathogenic exonic variant for 1 in 15 men, representing a life-long threat of transmission for these individuals and a significant burden on human population health.

Improving the clinical interpretation of missense variants in X linked genes using structural analysis

BACKGROUND

Improving the clinical interpretation of missense variants can increase the diagnostic yield of genomic testing and lead to personalised management strategies. Currently, due to the imprecision of bioinformatic tools that aim to predict variant pathogenicity, their role in clinical guidelines remains limited. There is a clear need for more accurate prediction algorithms and this study aims to improve performance by harnessing structural biology insights. The focus of this work is missense variants in a subset of genes associated with X linked disorders.

METHODS

We have developed a protein-specific variant interpreter (ProSper) that combines genetic and protein structural data. This algorithm predicts missense variant pathogenicity by applying machine learning approaches to the sequence and structural characteristics of variants.

RESULTS

ProSper outperformed seven previously described tools, including meta-predictors, in correctly evaluating whether or not variants are pathogenic; this was the case for 11 of the 21 genes associated with X linked disorders that met the inclusion criteria for this study. We also determined gene-specific pathogenicity thresholds that improved the performance of VEST4, REVEL and ClinPred, the three best-performing tools out of the seven that were evaluated; this was the case in 11, 11 and 12 different genes, respectively.

CONCLUSION

ProSper can form the basis of a molecule-specific prediction tool that can be implemented into diagnostic strategies. It can allow the accurate prioritisation of missense variants associated with X linked disorders, aiding precise and timely diagnosis. In addition, we demonstrate that gene-specific pathogenicity thresholds for a range of missense prioritisation tools can lead to an increase in prediction accuracy.

IDRMutPred: predicting disease-associated germline nonsynonymous single nucleotide variants (nsSNVs) in intrinsically disordered regions

Motivation

Despite of the lack of folded structure, intrinsically disordered regions (IDRs) of proteins play versatile roles in various biological processes, and many nonsynonymous single nucleotide variants (nsSNVs) in IDRs are associated with human diseases. The continuous accumulation of nsSNVs resulted from the wide application of NGS has driven the development of disease-association prediction methods for decades. However, their performance on nsSNVs in IDRs remains inferior, possibly due to the domination of nsSNVs from structured regions in training data. Therefore, it is highly demanding to build a disease-association predictor specifically for nsSNVs in IDRs with better performance.

Results

We present IDRMutPred, a machine learning-based tool specifically for predicting disease-associated germline nsSNVs in IDRs. Based on 17 selected optimal features that are extracted from sequence alignments, protein annotations, hydrophobicity indices and disorder scores, IDRMutPred was trained using three ensemble learning algorithms on the training dataset containing only IDR nsSNVs. The evaluation on the two testing datasets shows that all the three prediction models outperform 17 other popular general predictors significantly, achieving the ACC between 0.856 and 0.868 and MCC between 0.713 and 0.737. IDRMutPred will prioritize disease-associated IDR germline nsSNVs more reliably than general predictors.

Availability and implementation

The software is freely available at http://www.wdspdb.com/IDRMutPred.

Supplementary information

Supplementary data are available at Bioinformatics online.

MosaicBase: A Knowledgebase of Postzygotic Mosaic Variants in Noncancer Disease-related and Healthy Human Individuals

Mosaic variants resulting from postzygotic mutations are prevalent in the human genome and play important roles in human diseases. However, except for cancer-related variants, there is no collection of postzygotic mosaic variants in noncancer disease-related and healthy individuals. Here, we present MosaicBase, a comprehensive database that includes 6698 mosaic variants related to 266 noncancer diseases and 27,991 mosaic variants identified in 422 healthy individuals. Genomic and phenotypic information of each variant was manually extracted and curated from 383 publications. MosaicBase supports the query of variants with Online Mendelian Inheritance in Man (OMIM) entries, genomic coordinates, gene symbols, or Entrez IDs. We also provide an integrated genome browser for users to easily access mosaic variants and their related annotations for any genomic region. By analyzing the variants collected in MosaicBase, we find that mosaic variants that directly contribute to disease phenotype show features distinct from those of variants in individuals with mild or no phenotypes, in terms of their genomic distribution, mutation signatures, and fraction of mutant cells. MosaicBase will not only assist clinicians in genetic counseling and diagnosis but also provide a useful resource to understand the genomic baseline of postzygotic mutations in the general human population. MosaicBase is publicly available at http://mosaicbase.com/ or http://49.4.21.8:8000.

REVEL and BayesDel outperform other in silico meta-predictors for clinical variant classification

Many in silico predictors of genetic variant pathogenicity have been previously developed, but there is currently no standard application of these algorithms for variant assessment. Using 4,094 ClinVar-curated missense variants in clinically actionable genes, we evaluated the accuracy and yield of benign and deleterious evidence in 5 in silico meta-predictors, as well as agreement of SIFT and PolyPhen2, and report the derived thresholds for the best performing predictor(s). REVEL and BayesDel outperformed all other meta-predictors (CADD, MetaSVM, Eigen), with higher positive predictive value, comparable negative predictive value, higher yield, and greater overall prediction performance. Agreement of SIFT and PolyPhen2 resulted in slightly higher yield but lower overall prediction performance than REVEL or BayesDel. Our results support the use of gene-level rather than generalized thresholds, when gene-level thresholds can be estimated. Our results also support the use of 2-sided thresholds, which allow for uncertainty, rather than a single, binary cut-point for assigning benign and deleterious evidence. The gene-level 2-sided thresholds we derived for REVEL or BayesDel can be used to assess in silico evidence for missense variants in accordance with current classification guidelines.

Assessment of predicted enzymatic activity of α-N-acetylglucosaminidase variants of unknown significance for CAGI 2016

The NAGLU challenge of the fourth edition of the Critical Assessment of Genome Interpretation experiment (CAGI4) in 2016, invited participants to predict the impact of variants of unknown significance (VUS) on the enzymatic activity of the lysosomal hydrolase α-N-acetylglucosaminidase (NAGLU). Deficiencies in NAGLU activity lead to a rare, monogenic, recessive lysosomal storage disorder, Sanfilippo syndrome type B (MPS type IIIB). This challenge attracted 17 submissions from 10 groups. We observed that top models were able to predict the impact of missense mutations on enzymatic activity with Pearson's correlation coefficients of up to .61. We also observed that top methods were significantly more correlated with each other than they were with observed enzymatic activity values, which we believe speaks to the importance of sequence conservation across the different methods. Improved functional predictions on the VUS will help population-scale analysis of disease epidemiology and rare variant association analysis.

ATP1A3 mosaicism in families with alternating hemiplegia of childhood

Alternating hemiplegia of childhood (AHC) is a rare and severe neurodevelopmental disorder characterized by recurrent hemiplegic episodes. Most AHC cases are sporadic and caused by de novo ATP1A3 pathogenic variants. In this study, the aim was to identify the origin of ATP1A3 pathogenic variants in a Chinese cohort. In 105 probands including 101 sporadic and 4 familial cases, 98 patients with ATP1A3 pathogenic variants were identified, and 96.8% were confirmed as de novo. Micro-droplet digital polymerase chain reaction was applied for detecting ATP1A3 mosaicism in 80 available families. In blood samples, four asymptomatic parents, including two paternal and two maternal, and one proband with a milder phenotype were identified as mosaicism. Six (7.5%) parental mosaicisms were identified in multiple tissues, including four previously identified in blood and two additional cases identified from paternal sperms. Mosaicism was identified in multiple tissues with varied mutant allele fractions (MAFs, 0.03%-33.03%). The results suggested that MAF of mosaicism may be related to phenotype severity. This is the first systematic report of ATP1A3 mosaicism in AHC and showed mosaicism as an unrecognized source of previously considered "de novo" AHC. Identifying ATP1A3 mosaicism provides more evidence for estimating recurrence risk and has implications in genetic counseling of AHC.

A review study: Computational techniques for expecting the impact of non-synonymous single nucleotide variants in human diseases

Non-Synonymous Single-Nucleotide Variants (nsSNVs) and mutations can create a diversity effect on proteins as changing genotype and phenotype, which interrupts its stability. The alterations in the protein stability may cause diseases like cancer. Discovering of nsSNVs and mutations can be a useful tool for diagnosing the disease at a beginning stage. Many studies introduced the various predicting singular and consensus tools that based on different Machine Learning Techniques (MLTs) using diverse datasets. Therefore, we introduce the current comprehensive review of the most popular and recent unique tools that predict pathogenic variations and Meta-tool that merge some of them for enhancing their predictive power. Also, we scanned the several types computational techniques in the state-of-the-art and methods for predicting the effect both of coding and noncoding variants. We then displayed, the protein stability predictors. We offer the details of the most common benchmark database for variations including the main predictive features used by the different methods. Finally, we address the most common fundamental criteria for performance assessment of predictive tools. This review is targeted at bioinformaticians attentive in the characterization of regulatory variants, geneticists, molecular biologists attentive in understanding more about the nature and effective role of such variants from a functional point of views, and clinicians who may hope to learn about variants in human associated with a specific disease and find out what to do next to uncover how they impact on the underlying mechanisms.

A Bayesian framework for efficient and accurate variant prediction

There is a growing need to develop variant prediction tools capable of assessing a wide spectrum of evidence. We present a Bayesian framework that involves aggregating pathogenicity data across multiple in silico scores on a gene-by-gene basis and multiple evidence statistics in both quantitative and qualitative forms, and performs 5-tiered variant classification based on the resulting probability credible interval. When evaluated in 1,161 missense variants, our gene-specific in silico model-based meta-predictor yielded an area under the curve (AUC) of 96.0% and outperformed all other in silico predictors. Multifactorial model analysis incorporating all available evidence yielded 99.7% AUC, with 22.8% predicted as variants of uncertain significance (VUS). Use of only 3 auto-computed evidence statistics yielded 98.6% AUC with 56.0% predicted as VUS, which represented sufficient accuracy to rapidly assign a significant portion of VUS to clinically meaningful classifications. Collectively, our findings support the use of this framework to conduct large-scale variant prioritization using in silico predictors followed by variant prediction and classification with a high degree of predictive accuracy.

DeFine: deep convolutional neural networks accurately quantify intensities of transcription factor-DNA binding and facilitate evaluation of functional non-coding variants

The complex system of gene expression is regulated by the cell type-specific binding of transcription factors (TFs) to regulatory elements. Identifying variants that disrupt TF binding and lead to human diseases remains a great challenge. To address this, we implement sequence-based deep learning models that accurately predict the TF binding intensities to given DNA sequences. In addition to accurately classifying TF-DNA binding or unbinding, our models are capable of accurately predicting real-valued TF binding intensities by leveraging large-scale TF ChIP-seq data. The changes in the TF binding intensities between the altered sequence and the reference sequence reflect the degree of functional impact for the variant. This enables us to develop the tool DeFine (Deep learning based Functional impact of non-coding variants evaluator, http://define.cbi.pku.edu.cn) with improved performance for assessing the functional impact of non-coding variants including SNPs and indels. DeFine accurately identifies the causal functional non-coding variants from disease-associated variants in GWAS. DeFine is an effective and easy-to-use tool that facilities systematic prioritization of functional non-coding variants.

AutismKB 2.0: a knowledgebase for the genetic evidence of autism spectrum disorder

Autism spectrum disorder (ASD) is a complex neurodevelopmental disorder with strong genetic contributions. To provide a comprehensive resource for the genetic evidence of ASD, we have updated the Autism KnowledgeBase (AutismKB) to version 2.0. AutismKB 2.0 integrates multiscale genetic data on 1379 genes, 5420 copy number variations and structural variations, 11 669 single-nucleotide variations or small insertions/deletions (SNVs/indels) and 172 linkage regions. In particular, AutismKB 2.0 highlights 5669 de novo SNVs/indels due to their significant contribution to ASD genetics and includes 789 mosaic variants due to their recently discovered contributions to ASD pathogenesis. The genes and variants are annotated extensively with genetic evidence and clinical evidence. To help users fully understand the functional consequences of SNVs and small indels, we provided comprehensive predictions of pathogenicity with iFish, SIFT, Polyphen etc. To improve user experiences, the new version incorporates multiple query methods, including simple query, advanced query and batch query. It also functionally integrates two analytical tools to help users perform downstream analyses, including a gene ranking tool and an enrichment analysis tool, KOBAS. AutismKB 2.0 is freely available and can be a valuable resource for researchers.

Genomic mosaicism in paternal sperm and multiple parental tissues in a Dravet syndrome cohort

Genomic mosaicism in parental gametes and peripheral tissues is an important consideration for genetic counseling. We studied a Chinese cohort affected by a severe epileptic disorder, Dravet syndrome (DS). There were 56 fathers who donated semen and 15 parents who donated multiple peripheral tissue samples. We used an ultra-sensitive quantification method, micro-droplet digital PCR (mDDPCR), to detect parental mosaicism of the proband’s pathogenic mutation in SCN1A, the causal gene of DS in 112 families. Ten of the 56 paternal sperm samples were found to exhibit mosaicism of the proband’s mutations, with mutant allelic fractions (MAFs) ranging from 0.03% to 39.04%. MAFs in the mosaic fathers’ sperm were significantly higher than those in their blood (p = 0.00098), even after conditional probability correction (p’ = 0.033). In three mosaic fathers, ultra-low fractions of mosaicism (MAF < 1%) were detected in the sperm samples. In 44 of 45 cases, mosaicism was also observed in other parental peripheral tissues. Hierarchical clustering showed that MAFs measured in the paternal sperm, hair follicles and urine samples were clustered closest together. Milder epileptic phenotypes were more likely to be observed in mosaic parents (p = 3.006e-06). Our study provides new insights for genetic counseling.