Biology subjects

Gerstein, M. B.

Publications and source records attributed to Gerstein, M. B..

11 recordsLinked to original sources

bioRxiv2026

The dark transcriptome mouse organogenesis cell atlas

Single-cell atlases have transformed our understanding of mammalian tissue biology, yet the diverse noncoding "dark transcriptome" remains largely uncharted. Here, we present the Dark Transcriptome Mouse Organogenesis Cell Atlas (DT-MOCA), comprising 276,455 cells across embryonic days 11 to 18 and jointly capturing protein-coding transcripts and five major classes of noncoding RNAs through single-nucleus total RNA sequencing, establishing the first comprehensive catalog of noncoding RNAs across major cell types and developmental lineages. We found that long noncoding RNA and microRNA profiles distinguished major cell types and developmental lineages with resolution comparable to messenger RNA, whereas transfer RNA, small nuclear RNA, and small nucleolar RNA programs were less cell type-discriminative and more broadly shared. We further resolve noncoding RNA expression patterns as lineage-specific, temporally modulated, or coordinately regulated by cell identity and developmental stage. We observed that temporally regulated long noncoding RNAs were evolutionarily younger than those associated with cell identity. We further generate a catalog of the change of noncoding RNAs cross all 11 major developmental trajectories. It revealed lineage-specific coordination between transfer RNA supply and codon demand during musculoskeletal organogenesis as well as dynamic microRNA and small nucleolar RNA programs during the central nervous system development. We observed microRNAs converge on regulators of proliferation and gliogenesis and shared induction of Snord116 defining a developmental window relevant to Prader Willi syndrome. Together, this atlas provides a comprehensive catalog of the developmental dark transcriptome and reveals how noncoding RNA programs are organized across cell identity, developmental time, and lineage progression.

developmental biology↗

bioRxiv2023

Leveraging a large language model to predict protein phase transition: a physical, multiscale and interpretable approach

Protein phase transitions (PPTs) from the soluble state to a dense liquid phase (forming droplets via liquid-liquid phase separation) or to solid aggregates (such as amyloids) play key roles in pathological processes associated with age-related diseases such as Alzheimers disease. Several computational frameworks are capable of separately predicting the formation of droplets or amyloid aggregates based on protein sequences, yet none have tackled the prediction of both within a unified framework. Recently, large language models (LLMs) have exhibited great success in protein structure prediction; however, they have not yet been used for PPTs. Here, we fine-tune a LLM for predicting PPTs and demonstrate its usage in evaluating how sequence variants affect PPTs, an operation useful for protein design. In addition, we show its superior performance compared to suitable classical benchmarks. Due to the "black-box" nature of the LLM, we also employ a classical random forest model along with biophysical features to facilitate interpretation. Finally, focusing on Alzheimers disease-related proteins, we demonstrate that greater aggregation is associated with reduced gene expression in AD, suggesting a natural defense mechanism. Significance StatementProtein phase transition (PPT) is a physical mechanism associated with both physiological processes and age-related diseases. We present a modeling approach for predicting the protein propensity to undergo PPT, forming droplets or amyloids, directly from its sequence. We utilize a large language model (LLM) and demonstrate how variants within the protein sequence affect PPT. Because the LLM is naturally domain-agnostic, to enhance interpretability, we compare it with a classical knowledge-based model. Furthermore, our findings suggest the possible regulation of PPT by gene expression and transcription factors, hinting at potential targets for drug development. Our approach demonstrates the usefulness of fine-tuning a LLM for downstream tasks where only small datasets are available.

biophysics↗

bioRxiv2023

The ENCODE4 long-read RNA-seq collection reveals distinct classes of transcript structure diversity

The majority of mammalian genes encode multiple transcript isoforms that result from differential promoter use, changes in exonic splicing, and alternative 3 end choice. Detecting and quantifying transcript isoforms across tissues, cell types, and species has been extremely challenging because transcripts are much longer than the short reads normally used for RNA-seq. By contrast, long-read RNA-seq (LR-RNA-seq) gives the complete structure of most transcripts. We sequenced 264 LR-RNA-seq PacBio libraries totaling over 1 billion circular consensus reads (CCS) for 81 unique human and mouse samples. We detect at least one full-length transcript from 87.7% of annotated human protein coding genes and a total of 200,000 full-length transcripts, 40% of which have novel exon junction chains. To capture and compute on the three sources of transcript structure diversity, we introduce a gene and transcript annotation framework that uses triplets representing the transcript start site, exon junction chain, and transcript end site of each transcript. Using triplets in a simplex representation demonstrates how promoter selection, splice pattern, and 3 processing are deployed across human tissues, with nearly half of multitranscript protein coding genes showing a clear bias toward one of the three diversity mechanisms. Evaluated across samples, the predominantly expressed transcript changes for 74% of protein coding genes. In evolution, the human and mouse transcriptomes are globally similar in types of transcript structure diversity, yet among individual orthologous gene pairs, more than half (57.8%) show substantial differences in mechanism of diversification in matching tissues. This initial large-scale survey of human and mouse long-read transcriptomes provides a foundation for further analyses of alternative transcript usage, and is complemented by short-read and microRNA data on the same samples and by epigenome data elsewhere in the ENCODE4 collection.

genomics↗

bioRxiv2023

REPIC -- an ensemble learning methodology for cryo-EM particle picking

Cryo-EM (cryogenic electron microscopy) particle identification from micrographs (i.e., picking) is challenging due to the low signal-to-noise ratio and lack of ground truth for particle locations. Moreover, current computational methods ("pickers") identify different particle sets, complicating the selection of the best-suited picker for a protein of interest. Here, we present REPIC, an ensemble learning methodology that uses multiple pickers to find consensus particles. REPIC identifies consensus particles by framing its task as a graph problem and using integer linear programming to select particles. REPIC picks high-quality particles when the best picker is not known a priori and for known difficult-to-pick particles (e.g., TRPV1). Reconstructions using consensus particles achieve resolutions comparable to those from particles picked by experts, without the need for downstream particle filtering. Overall, our results show REPIC requires minimal (often no) manual picking and significantly reduces the burden on cryo-EM users for picker selection and particle picking. Availabilityhttps://github.com/ccameron/REPIC

bioinformatics↗

bioRxiv2022

Insights from Incorporating Quantum Computing into Drug Design Workflows

While many quantum computing (QC) methods promise theoretical advantages over classical counterparts, quantum hardware remains limited. Exploiting near-term QC in computer-aided drug design (CADD) thus requires judicious partitioning between classical and quantum calculations. We present HypaCADD, a hybrid classical-quantum workflow for finding ligands binding to proteins, while accounting for genetic mutations. We explicitly identify modules of our drug design workflow currently amenable to replacement by QC: non-intuitively, we identify the mutation-impact predictor as the best candidate. HypaCADD thus combines classical docking and molecular dynamics with quantum machine learning (QML) to infer the impact of mutations. We present a case study with the SARS-CoV-2 protease and associated mutants. We map a classical machine-learning module onto QC, using a neural network constructed from qubit-rotation gates. We have implemented this in simulation and on two commercial quantum computers. We find that the QML models can perform on par with, if not better than, classical baselines. In summary, HypaCADD offers a successful strategy for leveraging QC for CADD.

bioinformatics↗

bioRxiv2021

Nodal modulator is required to sustain endoplasmic reticulum morphology

Nodal Modulator (NOMO) is a widely conserved type I transmembrane protein of unknown function, with three nearly identical orthologs specified in the human genome. We identified NOMO1 in a proteomics approach aimed at the identification of proteins that support the structural integrity of the endoplasmic reticulum (ER). Overexpression of NOMO1 imposes a sheet morphology on the ER, while depletion of NOMO1 and its orthologs causes a collapse of ER morphology concomitant with the formation of membrane-delineated holes in the ER network. These structures are positive for the autophagy marker LAMP1, and LC3 is profoundly upregulated upon NOMO depletion. In vitro reconstitution of NOMO1 revealed a dimeric state that is mediated by the cytosolic tail domain, with each monomer featuring a "beads on a string" structure likely representing bacterial Ig-like folds. Based on these observations and a genetic epistasis analysis including the known ER-shaping proteins Atlastin2 and Climp63, we propose a role for NOMO1 in the functional network of ER-shaping proteins.

cell biology↗

bioRxiv2020

Expectations and blind spots for structural variation detection from short-read alignment and long-read assembly

Virtually all genome sequencing efforts in national biobanks, complex and Mendelian disease programs, and emerging clinical diagnostic approaches utilize short-reads (srWGS), which present constraints for genome-wide discovery of structural variants (SVs). Alternative long-read single molecule technologies (lrWGS) offer significant advantages for genome assembly and SV detection, while these technologies are currently cost prohibitive for large-scale disease studies and clinical diagnostics (∼5-12X higher cost than comparable coverage srWGS). Moreover, only dozens of such genomes are currently publicly accessible by comparison to millions of srWGS genomes that have been commissioned for international initiatives. Given this ubiquitous reliance on srWGS in human genetics and genomics, we sought to characterize and quantify the properties of SVs accessible to both srWGS and lrWGS to establish benchmarks and expectations in ongoing medical and population genetic studies, and to project the added value of SVs uniquely accessible to each technology. In analyses of three trios with matched srWGS and lrWGS from the Human Genome Structural Variation Consortium (HGSVC), srWGS captured ∼11,000 SVs per genome using reference-based algorithms, while haplotype-resolved assembly from lrWGS identified ∼25,000 SVs per genome. Detection power and precision for SV discovery varied dramatically by genomic context and variant class: 9.7% of the current GRCh38 reference is defined by segmental duplications (SD) and simple repeats (SR), yet 91.4% of deletions that were specifically discovered by lrWGS localized to these regions. Across the remaining 90.3% of the human reference, we observed extremely high concordance (93.8%) for deletions discovered by srWGS and lrWGS after error correction using the raw lrWGS reads. Conversely, lrWGS was superior for detection of insertions across all genomic contexts. Given that the non-SD/SR sequences span 90.3% of the GRCh38 reference, and encompass 95.9% of coding exons in currently annotated disease associated genes, improved sensitivity from lrWGS to discover novel and interpretable pathogenic deletions not already accessible to srWGS is likely to be incremental. However, these analyses highlight the added value of assembly-based lrWGS to create new catalogues of functional insertions and transposable elements, as well as disease associated repeat expansions in genomic regions previously recalcitrant to routine assessment.Competing Interest StatementThe authors have declared no competing interest.View Full Text

genomics↗

bioRxiv2019

LESSeq: Local event-based analysis of alternative splicing using RNA-Seq data

Alternative splicing, which can be observed genome-wide by RNA-Seq, is important in cellular development and evolution. Comparative RNA-Seq experiments between different cellular conditions allow alternative splicing signatures to be detected. However, inferring alternative splicing signatures from short-read technology is unreliable and still presents many challenges before biologically significant signatures may be identified. To enable the robust discovery of differential alternative splicing, we developed the Local Event-based analysis of alternative Splicing using RNA-Seq (LESSeq) pipeline. LESSeq utilizes information of local splicing events (i.e., the partial structures in genes where transcript-splicing patterns diverge) to identify unambiguous alternative splicing. In addition, LESSeq quantifies the abundance of these alternative events using Maximum Likelihood Estimation (MLE) and provides their significance between different cellular conditions. The utility of LESSeq is demonstrated through two case studies relevant to human variation and evolution. Using an RNA-Seq data set of lymphoblastoid cell lines in two human populations, we examined within-species variation and discovered population-differential alternative splicing events. With an RNA-Seq data set of several tissues in human and rhesus macaque, we studied cross-species variation and identified lineage-differential alternative splicing events. LESSeq is implemented in C++ and R, and made publicly available on GitHub at: https://github.com/gersteinlab/LESSeq

bioinformatics↗

bioRxiv2019

Latent-space embedding of expression data identifies gene signatures from sputum samples of asthmatic patients

The pathogenesis of asthma is a complex process involving multiple genes and pathways. Identifying biomarkers from asthma datasets, especially those that include heterogeneous subpopulations, is challenging. In this work, we developed a framework that incorporates a denoising autoencoder and a supervised learning approach to identify gene signatures related to asthma severity. The autoencoder embeds high-dimensional gene expression data into a lower-dimensional latent space in an unsupervised fashion, enabling us to extract distinguishing features from gene expression data. We found that the weights on hidden units in this latent space correlate well with previously defined and clinically relevant clusters of patients. Moreover, pathway analysis based on each genes contribution to the hidden units showed significant enrichment in known asthma-related pathways. We then used genes that contribute most to the hidden units to develop a secondary supervised classifier (based on random forest) for directly predicting asthma severity. The random-forest importance metric from this classifier identified a signature based on 50 key genes, which can predict severity with an AUROC of 0.81 and thus have potential as diagnostic biomarkers. Furthermore, the key genes could also be used for successfully estimating, via support-vector-machine regression, the FEV1/FVC ratios across patients, achieving pre- and post-treatment correlations of 0.56 and 0.65, respectively (between predicted and observed values). The 50 biomarker candidate genes can be found in supplementary. The source codes are freely available upon request.

bioinformatics↗

bioRxiv2018

Genomic variants concurrently listed in a somatic and a germline mutation database have implications for disease-variant discovery and genomic privacy

BackgroundMutations arise in the human genome in two major settings: the germline and soma. These settings involve different inheritance patterns, chromatin structures, and environmental exposures, all of which might be predicted to differentially affect the distribution of substitutions found in these settings. Nonetheless, recent studies have found that somatic and germline mutation rates are similarly affected by endogenous mutational processes and epigenetic factors.\n\nResultsHere, we quantified the number of single nucleotide variants that co-occur between somatic and germline call-sets (cSNVs), compared this quantity with expectations, and explained noted departures. We found that three times as many variants are shared between the soma and germline than is expected by independence. We developed a new, general-purpose statistical framework to explain the observed excess of cSNVs in terms of the varying mutation rates of different kinds substitution types and of genomic regions. Using this metric, we find that more than 90% of this excess can be explained by our observation that the basic substitution types (such as N[C->T]G, C->A, etc.) have correlated mutation rates in the germline and soma. Matched-normal read depth analysis suggests that an appreciable fraction of this excess may also derive from germline contamination of somatic samples.\n\nConclusionOverall, our results highlight the commonalities in substitution patterns between the germline and soma. The universality of some aspects of human mutation rates offers insight into the potential molecular mechanisms of human mutation. The highlighted similarities between somatic and germline mutation rates also lay the groundwork for future studies that distinguish disease-causing variants from a genomic background informed by both somatic and germline variant data. Moreover, our results also indicate that the depth of matched normal sequencing necessary to ensure genomic privacy of donors of somatic samples may be higher than previously appreciated. Furthermore, the fact that we were able to explain such a high portion of recurrent variants using known determinants of mutation rates is evidence that the genomics community has already discovered the most important predictors of mutation rates for single nucleotide variants.

genomics↗

bioRxiv2018

Reliability of Whole-Exome Sequencing for Assessing Intratumor Genetic Heterogeneity

Multi-region sequencing is used to detect intratumor genetic heterogeneity (ITGH) in tumors. To assess whether genuine ITGH can be distinguished from sequencing artifacts, we whole-exome sequenced (WES) three anatomically distinct regions of the same tumor with technical replicates to estimate technical noise. Somatic variants were detected with three different WES pipelines and subsequently validated by high-depth amplicon sequencing. The cancer-only pipeline was unreliable, with about 69% of the identified somatic variants being false positive. Even with matched normal DNA where 82% of the somatic variants were detected reliably, only 36%-78% were found consistently in technical replicate pairs. Overall 34%-80% of the discordant somatic variants, which could be interpreted as ITGH, were found to constitute technical noise. Excluding mutations affecting low mappability regions or occurring in certain mutational contexts was found to reduce artifacts, yet detection of subclonal mutations by WES in the absence of orthogonal validation remains unreliable.

genomics↗