Biology subjects

Singh, K. K.

Publications and source records attributed to Singh, K. K..

4 recordsLinked to original sources

bioRxiv2018

Inference of splicing motifs through visualization of recurrent networks

Neural models have been able to obtain state-of-the-art performances on several genome sequence-based prediction tasks. Such models take only nucleotide sequences as input and learn relevant features on its own. However, extracting the interpretable motifs from the model remains a challenge. This work explores various existing visualization techniques in their ability to infer relevant sequence information learned by a recurrent neural network (RNN) on the task of splice junction identification. The visualization techniques have been modulated to suit the genome sequences as input. The visualizations inspect genomic regions at the level of a single nucleotide as well as a span of consecutive nucleotides. This inspection is performed based on modification of input sequences (perturbation-based) or the embedding space (back-propagation based). We infer features pertaining to both canonical and non-canonical splicing from a single neural model. Results indicate that the visualization techniques produce comparable performance for branchpoint detection. However, in case of canonical donor and acceptor junction motifs, perturbation based visualizations perform better than back-propagation based visualizations and vice-versa for non-canonical motifs.

bioinformatics↗

bioRxiv2018

CNTN5-/+ or EHMT2-/+ iPSC-Derived Neurons from Individuals with Autism Develop Hyperactive Neuronal Networks

Induced pluripotent stem cell (iPSC)-derived cortical neurons are increasingly used as a model to study developmental aspects of Autism Spectrum Disorder (ASD), which is clinically and genetically heterogeneous. To study the complex relationship of rare (penetrant) variant(s) and common (weaker) polygenic risk variant(s) to ASD, \"isogenic\" iPSC-derived neurons from probands and family-based controls, for modeling, is critical. We developed a standardized set of procedures, designed to control for heterogeneity in reprogramming and differentiation, and generated 53 different iPSC-derived glutamatergic neuronal lines from 25 participants from 12 unrelated families with ASD (14 ASD-affected individuals, 3 unaffected siblings, 8 unaffected parents). Heterozygous de novo (7 families; 16p11.2, NRXN1, DLGAP2, CAPRIN1, VIP, ANOS1, THRA) and rare-inherited (2 families; CNTN5, AGBL4) presumed-damaging variants were characterized in ASD risk genes/loci. In three additional families, functional candidates for ASD (SET), and combinations of putative etiologic variants (GLI3/KIF21A and EHMT2/UBE2I combinations in separate families), were modeled. We used a large-scale multi-electrode array (MEA) as our primary high-throughput phenotyping assay, followed by patch clamp recordings. Our most compelling new results revealed a consistent spontaneous network hyperactivity in neurons deficient for CNTN5 or EHMT2. Our biobank of iPSC-derived neurons and accompanying genomic data are available to accelerate ASD research.

neuroscience↗

bioRxiv2018

Complete Disruption of Autism-Susceptibility Genes by Gene-Editing Predominantly Reduces Functional Connectivity of Isogenic Human Neurons

Autism Spectrum Disorder is phenotypically and genetically heterogeneous, but genomic analyses have identified candidate susceptibility genes. We present a CRISPR gene editing strategy to insert a protein tag and premature termination sites creating an induced pluripotent stem cell (iPSC) knockout resource for functional studies of 10 ASD-relevant genes (AFF2/FMR2, ANOS1, ASTN2, ATRX, CACNA1C, CHD8, DLGAP2, KCNQ2, SCN2A, TENM1). Neurogenin 2 (NEUROG2)-directed differentiation of iPSCs allowed production of cortical excitatory neurons, and mutant proteins were not detectable. RNAseq revealed convergence of several neuronal networks. Using both patch-clamp and multi-electrode array approaches, the electrophysiological deficits measured were distinct for different mutations. However, they culminated in a consistent reduction in synaptic activity, including reduced spontaneous excitatory post-synaptic current frequencies in AFF2/FMR2-, ASTN2-, ATRX-, KCNQ2- and SCN2A-null neurons. Despite ASD susceptibility genes belonging to different gene ontologies, isogenic stem cell resources can reveal common functional phenotypes, such as reduced functional connectivity.

neuroscience↗

bioRxiv2017

SpliceVec: distributed feature representations for splice junction prediction

Identification of intron boundaries, called splice junctions, is an important part of delineating gene structure and functions. This also provides valuable in-sights into the role of alternative splicing in increasing functional diversity of genes. Identification of splice junctions through RNA-seq is by mapping short reads to the reference genome which is prone to errors due to random sequence matches. This encourages identification of splicing junctions through computa-tional methods based on machine learning. Existing models are dependent on feature extraction and selection for capturing splicing signals lying in vicinity of splice junctions. But such manually extracted features are not exhaustive. We introduce distributed feature representation, SpliceVec, to avoid explicit and biased feature extraction generally adopted for such tasks. SpliceVec is based on two widely used distributed representation models in natural language processing. Learned feature representation in form of SpliceVec is fed to multi-layer perceptron for splice junction classification task. An intrinsic evaluation of SpliceVec indicates that it is able to group true and false sites distinctly. Our study on optimal context to be considered for feature extraction indicates inclusion of entire intronic sequence to be better than flanking upstream and downstream region around splice junctions. Further, SpliceVec is invariant to canonical and non-canonical splice junction detection. The proposed model is consistent in its performance even with reduced dataset and class-imbalanced dataset. SpliceVec is computationally efficient and can be trained with user defined data as well.

bioinformatics↗