Biology subjects

Rabiee, H. R.

Publications and source records attributed to Rabiee, H. R..

6 recordsLinked to original sources

bioRxiv2022

DeepGenePrior: A deep learning model to prioritize genes affected by copy number variants

The genetic etiology of neurodevelopmental disorders is highly heterogeneous. They are characterized by abnormalities in the development of the central nervous system, which lead to diminished physical or intellectual capabilities. Determining which gene is the driver of disease (not just a passenger), termed gene prioritization, is not entirely known. In terms of disease-gene associations, genome-wide explorations are still underdeveloped due to the reliance on previous discoveries when spotting new genes and other evidence sources with false positive or false negative relations. This paper introduces DeepGenePrior, a model based on deep neural networks that prioritizes candidate genes in Copy Number Variant (CNV) mediated diseases. Based on the well-studied Variational AutoEncoder (VAE), we developed a score to measure the impact of the genes on the target diseases. Unlike other methods that use prior data on gene-disease associations to prioritize candidate genes (using the guilt by association principle), the current study exclusively relies on copy number variants. Therefore, the procedure can identify disease-associated genes regardless of prior knowledge or auxiliary data sources. We identified genes that distinguish cases from disorders (autism, schizophrenia, and developmental delay). A 12% increase in fold enrichment was observed in brain-expressed genes compared to previous studies, while 15% more fold enrichment was found in genes associated with mouse nervous system phenotypes. We also explored sex dimorphism for the disorders and discovered genes that overexpress more in one gender than the other. Additionally, we investigated the gene ontology of the putative genes with WebGestalt and the associations between the causative genes and the other phenotypes in the DECIPHER dataset. Furthermore, some genes were jointly present in the top genes associated with the three disorders in this study (i.e., autism spectrum disorder, schizophrenia, and developmental delay); namely, deletions in ZDHHC8, DGCR5, and CATG00000022283 were common between them. These findings suggest the common etiology of these clinically distinct conditions. With DeepGenePrior, we address the obstacles in existing gene prioritization studies. This study identified promising candidate genes without prior knowledge of diseases or phenotypes using deep learning.

bioinformatics↗

bioRxiv2021

Somatic point mutations are enriched in long non-coding RNAs with possible regulatory function in breast cancer

De novo somatic point mutations identified in breast cancer are predominantly non-coding and typically attributed to altered regulatory elements such as enhancers and promoters. However, while the non-coding RNAs (ncRNAs) form a large portion of the mammalian genome, their biological functions are mostly poorly characterized in cancers. In this study, using a newly developed tool, SomaGene, we reanalyze de novo somatic point mutations from the International Cancer Genome Consortium (ICGC) whole-genome sequencing data of 1,855 breast cancers. We identify 929 candidates of ncRNAs that are significantly and explicitly mutated in breast cancer samples. By integrating data from the ENCODE regulatory features and FANTOM5 expression atlas, we show that the candidate ncRNAs in breast cancer samples significantly enrich for active chromatin histone marks (1.9 times), CTCF binding sites (2.45 times), DNase accessibility (1.76 times), HMM predicted enhancers (2.26 times) and eQTL polymorphisms (1.77 times). Importantly, we show that the 929 ncRNAs contain a much higher level (3.64 times) of breast cancer-associated genome-wide association (GWAS) single nucleotide polymorphisms (SNPs) than genome-wide expectation. Such enrichment has not been seen with GWAS SNPs from other diseases. Using breast tissue related Hi-C data we then show that 82% of our candidate ncRNAs (1.9 times) significantly interact with the promoter of protein-coding genes, including previously known cancer-associated genes, suggesting the critical role for candidate ncRNA genes in activation of essential regulators of development and differentiation in breast cancer. We provide an extensive web-based resource (http://ncrna.ictic.sharif.edu), to communicate our results with the research community. Our list of breast cancer-specific ncRNA genes has the potential to provide a better understanding of the underlying genetic causes of breast cancer. Lastly, the tool developed in this study can be used in the analysis of somatic mutations in all cancers.

cancer biology↗

bioRxiv2020

Integrative analysis of mutated genes and mutational processes reveals seven colorectal cancer subtypes

Colorectal cancer (CRC) is one of the leading causes of cancer-related deaths in the world. It has been reported that [~]10%-15% of individuals with colorectal cancer experience a causative mutation in the known susceptibility genes, highlighting the importance of identifying mutations for early detection in high risk individuals. Through extensive sequencing projects such as the International Cancer Genome Consortium (ICGC), a large number of somatic point mutations have been identified that can be used to identify cancer-associated genes, as well as the signature of mutational processes defined by the tri-nucleotide sequence context (motif) of mutated sites. Mutation is the hallmark of cancer genome, and many studies have reported cancer subtyping based on the type of frequently mutated genes, or the proportion of mutational processes, however, none of these cancer subtyping methods consider these features simultaneously. This highlights the need for a better and more inclusive subtype classification approach to enable biomarker discovery and thus inform drug development for CRC. In this study, we developed a statistical pipeline based on a novel concept gene-motif, which merges mutated gene information with tri-nucleotide motif of mutated sites, to identify cancer subtypes, in this case CRCs. Our analysis identified for the first time, 3,131 gene-motif combinations that were significantly mutated in 536 ICGC colorectal cancer samples compared to other cancer types, identifying seven CRC subtypes with distinguishable phenotypes and biomarkers. Interestingly, we identified several genes that were mutated in multiple subtypes but with unique sequence contexts. Taken together, our results highlight the importance of considering both the mutation type and mutated genes in identification of cancer subtypes and cancer biomarkers.

cancer biology↗

bioRxiv2020

MaxHiC: robust estimation of chromatin interaction frequency in Hi-C and capture Hi-C experiments

Hi-C is a genome-wide chromosome conformation capture technology that detects interactions between pairs of genomic regions, and exploits higher order chromatin structures. Conceptually Hi-C data counts interaction frequencies between every position in the genome and every other position. Biologically functional interactions are expected to occur more frequently than random (background) interactions. To identify biologically relevant interactions, several background models that take biases such as distance, GC content and mappability into account have been proposed. Here we introduce MaxHiC, a background correction tool that deals with these complex biases and robustly identifies statistically significant interactions in both Hi-C and capture Hi-C experiments. MaxHiC uses a negative binomial distribution model and a maximum likelihood technique to correct biases in both Hi-C and capture Hi-C libraries. We systematically benchmark MaxHiC against major Hi-C background correction tools and demonstrate using published Hi-C and capture Hi-C datasets that 1) Interacting regions identified by MaxHiC have significantly greater levels of overlap with known regulatory features (e.g. active chromatin histone marks, CTCF binding sites, DNase sensitivity) and also disease-associated genome-wide association SNPs than those identified by currently existing models, and 2) the pairs of interacting regions are more likely to be linked by eQTL pairs and more likely to link known regulatory features than any of the existing methods. We also demonstrate that interactions between different genomic region types have distinct distance distribution only revealed by MaxHiC. MaxHiC is publicly available as a python package for the analysis of Hi-C and capture Hi-C data. Author summaryMaxHiC is a robust machine learning based tool for identifying significant interacting regions from both Hi-C and capture Hi-C data. All the current existing models are designed for either Hi-C or capture Hi-C data, however we developed MaxHiC to be applicable for both Hi-C and capture Hi-C libraries (two different models have been used for Hi-C and capture Hi-C data). MaxHiC is also able to analyse very deep Hi-C libraries (e.g., MicroC) without any computational issues. MaxHiC significantly outperforms current existing tools in terms of enrichment of interactions between known regulatory regions as well as biologically relevant interactions.

bioinformatics↗

bioRxiv2018

CANCERSIGN: a user-friendly and robust tool for identification and classification of mutational signatures and patterns in cancer genomes

Analyses of large somatic mutation datasets, using advanced computational algorithms, have revealed at least 30 independent mutational signatures in tumor samples. These studies have been instrumental in identification and quantification of responsible endogenous and exogenous molecular processes against cancer. The quantitative approach used to deconvolute mutational signatures is becoming an integral part of cancer research. Therefore, development of a stand-alone tool with a user-friendly graphical interface for analysis of cancer mutational signatures is necessary. In this manuscript, we introduce CANCERSIGN as an open access1 bioinformatics tool that uses raw mutation data (BED files) as input, and generates 3-mer and 5-mer mutational signatures. Additionally, this tool enables users to perform clustering on tumor samples based on the raw mutation counts as well as using the proportion of mutational signatures in each sample. Using this tool, we analysed all the whole genome somatic mutation datasets of International Cancer Genome Consortium (ICGC) samples and identified a number of novel signatures.

bioinformatics↗

bioRxiv2018

Deep Learning Based Proarrhythmia Analysis Using Field Potentials Recorded from Human Pluripotent Stem Cells Derived Cardiomyocytes

AimAn early characterization of drug-induced cardiotoxicity may be possible by combining comprehensive in vitro pro-arrhythmia assay and deep learning techniques. The goal of this study was to develop a deep learning method to automatically detect irregular beating rhythm as well as abnormal waveforms of field potentials in an in vitro cardiotoxicity assay using human pluripotent stem cell (hPSC) derived cardiomyocytes and multi-electrode array (MEA) system.\n\nMethods and ResultsWe included field potential waveforms from 380 experiments which obtained by application of some cardioactive drugs on healthy and/or patient-specific induced pluripotent stem cells derived cardiomyocytes (iPSC-CM). We employed convolutional and recurrent neural networks, in order to develop a new method for automatic classification of field potential recordings without using any hand-engineered features. In the proposed method, a preparation phase was initially applied to split 60-second long recordings into a series of 5-second long windows. Thereafter, the classification phase comprising of two main steps was designed. In the first step, 5-second long windows were classified using a designated convolutional neural network (CNN). In the second step, the results of 5-second long window assessments were used as the input sequence to a recurrent neural network (RNN). The output was then compared to electrophysiologist-level arrhythmia (irregularity or abnormal waveforms) detection, resulting in 0.84 accuracy, 0.84 sensitivity, 0.85 specificity, and 0.88 precision.\n\nConclusionA novel deep learning approach based on a two-step CNN-RNN method can be used for automated analysis of \"irregularity or abnormal waveforms\" in an in vitro model of cardiotoxicity experiments.

bioinformatics↗