Scholars@Duke
Rohit Singh

Scholarly Works

COMPASS: Component-Wise Inference of Shared and Gene-Specific Perturbation Response

Preprint · August 6, 2026 Predicting how a genetic perturbation reshapes a cell's transcriptome is a central goal of computational biology. Previous studies report that the mean response across training perturbations rivals specialized models on standard accuracy metrics, e ... Full text Cite

Miniaturizing and modifying natural proteins with Raygun.

Journal article Nature · July 29, 2026 Proteins have evolved over billions of years through coordinated substitutions, insertions and deletions, yet computational protein design cannot fully replicate nature's ability to engineer new proteins from existing templates. Protein language models1-3 ... Full text Link to item Cite

Foundation model reveals the shared organization of transcription and topologically associating domains.

Journal article Cell Syst · July 17, 2026 The three-dimensional organization of chromatin into topologically associating domains (TADs) may impact gene regulation by bringing distant genes into contact. However, studies of TADs' function and their influence on transcription have been constrained b ... Full text Link to item Cite

Learning a PRECISE language for small-molecule binding

Preprint · 2026 Virtual screening of billion-scale compound libraries has become feasible through machine learning approaches. In particular, CoNCISE (RECOMB 2025) introduced drug quantization via code-books, achieving highly scalable and accurate binary predictions. Howe ... Full text Cite

EvoPool: Evolution-Guided Pooling of Protein Language Model Embeddings

Preprint · 2026 Protein language models (PLMs) encode amino acid sequences into residue-level embeddings that must be pooled into fixed-size representations for downstream protein-level prediction tasks. Although these embeddings implicitly reflect evolutionary constraint ... Full text Cite

DeltaNMF: A Two-Stage Neural NMF for Differential Gene Program Discovery

Preprint · 2026 Non-negative matrix factorization (NMF) is a foundational dimensionality-reduction method in single-cell transcriptomics, valued for its interpretable gene programs. However, in case-control settings common in perturbation and disease research, standard NM ... Full text Cite

TP53-mediated bidirectional lineage plasticity drives alveolar epithelial cell extrusion and tissue remodeling

Preprint · 2026 Cell extrusion contributes to epithelial homeostasis, but its dysregulation can lead to tumorigenesis or degeneration. A fine balance in this process is therefore essential for tissue integrity. Yet the cell types and states vulnerable to extrusion, and th ... Full text Cite

Democratizing protein language model training, sharing and collaboration.

Journal article Nat Biotechnol · October 24, 2025 Training and deploying large-scale protein language models typically requires deep machine learning expertise-a barrier for researchers outside this field. SaprotHub overcomes this challenge by offering an intuitive platform that facilitates training and p ... Full text Link to item Cite

Unveiling causal regulatory mechanisms through cell-state parallax.

Journal article Nat Commun · August 29, 2025 Genome-wide association studies (GWAS) identify numerous disease-linked genetic variants at noncoding genomic loci, yet therapeutic progress is hampered by the challenge of deciphering the regulatory roles of these loci in tissue-specific contexts. Single- ... Full text Link to item Cite

The basic helix-loop-helix transcription factor TCF4 recruits the Mediator Complex to activate gonadal genes and drive ovarian development.

Journal article bioRxiv · July 11, 2025 The bipotential gonad is the precursor organ to both the ovary and testis and develops as part of the embryonic urogenital system. In mice, gonadogenesis initiates around embryonic day 9.5 (E9.5), when coelomic epithelial (CE) cells overlaying the mesoneph ... Full text Link to item Cite

Comparative analysis of the syncytiotrophoblast in placenta tissue and trophoblast organoids using snRNA sequencing.

Preprint · May 27, 2025 The syncytiotrophoblast (STB) is a multinucleated cell layer that forms the outer surface of human chorionic villi. Its unusual structure, with billions of nuclei in a single cell, makes it difficult to resolve using conventional single-cell methods. To be ... Full text Link to item Cite

Comparative analysis of the syncytiotrophoblast in placenta tissue and trophoblast organoids using snRNA sequencing

Journal article eLife · May 27, 2025 The syncytiotrophoblast (STB) is a multinucleated cell layer that forms the outer surface of human chorionic villi. Its unusual structure, with billions of nuclei in a single cell, makes it difficult to resolve using conventional single-cell method ... Full text Cite

Tracing the Shared Foundations of Gene Expression and Chromatin Structure.

Preprint · April 2, 2025 UNLABELLED: The three-dimensional organization of chromatin into topologically associating domains (TADs) may impact gene regulation by bringing distant genes into contact. However, many questions about TADs' function and their influence on transcription r ... Full text Link to item Cite

Topology-driven discovery of transmembrane protein S-palmitoylation.

Journal article J Biol Chem · March 2025 Protein S-palmitoylation is a reversible lipophilic posttranslational modification regulating diverse signaling pathways. Within transmembrane proteins (TMPs), S-palmitoylation is implicated in conditions from inflammatory disorders to respiratory viral in ... Full text Link to item Cite

Comparative analysis of the syncytiotrophoblast in placenta tissue and trophoblast organoids using snRNA sequencing.

Journal article bioRxiv · February 19, 2025 The outer surface of chorionic villi in the human placenta consists of a single multinucleated cell called the syncytiotrophoblast (STB). The unique cellular ultrastructure of the STB presents challenges in deciphering its gene expression signature at the ... Full text Link to item Cite

Learning the language of antibody hypervariability.

Journal article Proc Natl Acad Sci U S A · January 7, 2025 Protein language models (PLMs) have demonstrated impressive success in modeling proteins. However, general-purpose "foundational" PLMs have limited performance in modeling antibodies due to the latter's hypervariable regions, which do not conform to the ev ... Full text Link to item Cite

Learning a CoNCISE language for small-molecule binding

Preprint · 2025 Rapid advances in deep learning have improved in silico methods for drug-target interaction (DTI) prediction. However, current methods do not scale to the massive catalogs that list millions or billions of commercially-available small molecules. Here, we ... Full text Cite

Aggregating residue-level protein language model embeddings with optimal transport.

Journal article Bioinform Adv · 2025 MOTIVATION: Protein language models (PLMs) have emerged as powerful approaches for mapping protein sequences into embeddings suitable for various applications. As protein representation schemes, PLMs generate per-token (i.e. per-residue) representations, r ... Full text Open Access Link to item Cite

Learning a CoNCISE Language for Small-Molecule Binding

Book section · January 1, 2025 Rapid advances in deep learning have improved in silico methods for drug-target interaction (DTI) prediction. However, current methods struggle to scale to catalogs listing billions of commercially-available small molecules. Here, we introduce CoNCISE, a m ... Full text Cite

Decoding the Functional Interactome of Non-model Organisms with PHILHARMONIC

Conference Lecture Notes in Computer Science · January 1, 2025 We introduce PHILHARMONIC, a computational framework that couples deep learning de novo network inference with robust unsupervised spectral clustering algorithms to uncover functional relationships and high-level organization in non-model organisms. Our no ... Full text Cite

SAME: Topology-flexible transforms enable robust integration of multimodal spatial omics

Preprint · 2025 Spatial omics technologies provide complementary and layered molecular insights that span proteins, transcripts, and metabolites. However, aligning and integrating these modalities across serial tissue sections remains a computational challenge. Existing a ... Full text Cite

USHER: Guiding Foundation Model Representations through Distribution Shifts

Preprint · 2025 Foundation models pre-trained on certain biological data modalities exhibit systematic representational biases when encountering out-of-distribution (OOD) data from new assays. The embedding drift largely arises from instrumentation and protocol-related ar ... Full text Cite

Topology-Driven Discovery of Transmembrane Protein S-Palmitoylation.

Preprint · September 8, 2024 Protein S-palmitoylation is a reversible lipophilic posttranslational modification regulating a diverse number of signaling pathways. Within transmembrane proteins (TMPs), S-palmitoylation is implicated in conditions from inflammatory disorders to respirat ... Full text Link to item Cite

Causal gene regulatory analysis with RNA velocity reveals an interplay between slow and fast transcription factors.

Journal article Cell Syst · May 15, 2024 Single-cell expression dynamics, from differentiation trajectories or RNA velocity, have the potential to reveal causal links between transcription factors (TFs) and their target genes in gene regulatory networks (GRNs). However, existing methods either ov ... Full text Link to item Cite

Miniaturizing, Modifying, and Magnifying Nature’s Proteins with Raygun

Preprint · 2024 Proteins have evolved over billions of years through extensive and coordinated substitutions, insertions and deletions (indels). Computational protein design cannot yet fully mimic nature’s ability to engineer new proteins from existing templates. Protein ... Full text Cite

TT3D: Leveraging precomputed protein 3D sequence models to predict protein-protein interactions.

Journal article Bioinformatics · November 1, 2023 MOTIVATION: High-quality computational structural models are now precomputed and available for nearly every protein in UniProt. However, the best way to leverage these models to predict which pairs of proteins interact in a high-throughput manner is not im ... Full text Link to item Cite

Contrastive learning in protein language space predicts interactions between drugs and protein targets.

Journal article Proc Natl Acad Sci U S A · June 13, 2023 Sequence-based prediction of drug-target interactions has the potential to accelerate drug discovery by complementing experimental screens. Such computational prediction needs to be generalizable and scalable while remaining sensitive to subtle variations ... Full text Link to item Cite

split-intein Gal4 provides intersectional genetic labeling that is repressible by Gal80.

Journal article Proc Natl Acad Sci U S A · June 13, 2023 The split-Gal4 system allows for intersectional genetic labeling of highly specific cell types and tissues in Drosophila. However, the existing split-Gal4 system, unlike the standard Gal4 system, cannot be repressed by Gal80, and therefore cannot be contro ... Full text Link to item Cite

Transfer of knowledge from model organisms to evolutionarily distant non-model organisms: The coral Pocillopora damicornis membrane signaling receptome.

Journal article PLoS One · 2023 With the ease of gene sequencing and the technology available to study and manipulate non-model organisms, the extension of the methodological toolbox required to translate our understanding of model organisms to non-model organisms has become an urgent pr ... Full text Link to item Cite

Local transcriptional covariation produces accurate estimates of cell phenotype

Preprint · 2023 The utility of single-cell RNA sequencing (scRNA-seq) is premised on the notion that transcriptional state can faithfully reflect cell phenotype. However, scRNA-seq measurements are noisy and sparse, with individual transcript counts showing limited correl ... Full text Cite

Causally-guided Regularization of Graph Attention Improves Generalizability

Journal article Transactions on Machine Learning Research · January 1, 2023 Graph attention networks estimate the relational importance of node neighbors to aggregate relevant information over local neighborhoods for a prediction task. However, the inferred attentions are vulnerable to spurious correlations and connectivity in the ... Cite

Topsy-Turvy: integrating a global view into sequence-based PPI prediction.

Journal article Bioinformatics · June 24, 2022 SUMMARY: Computational methods to predict protein-protein interaction (PPI) typically segregate into sequence-based 'bottom-up' methods that infer properties from the characteristics of the individual protein sequences, or global 'top-down' methods that in ... Full text Link to item Cite

GRANGER CAUSAL INFERENCE ON DAGS IDENTIFIES GENOMIC LOCI REGULATING TRANSCRIPTION

Conference Iclr 2022 10th International Conference on Learning Representations · January 1, 2022 When a dynamical system can be modeled as a sequence of observations, Granger causality is a powerful approach for detecting predictive interactions between its variables. However, traditional Granger causal inference has limited utility in domains where t ... Cite

D-SCRIPT translates genome to phenome with sequence-based, structure-aware, genome-scale predictions of protein-protein interactions.

Journal article Cell Syst · October 20, 2021 We combine advances in neural language modeling and structurally motivated design to develop D-SCRIPT, an interpretable and generalizable deep-learning model, which predicts interaction between two proteins using only their sequence and maintains high accu ... Full text Link to item Cite

Schema: metric learning enables interpretable synthesis of heterogeneous single-cell modalities.

Journal article Genome Biol · May 3, 2021 A complete understanding of biological processes requires synthesizing information across heterogeneous modalities, such as age, disease status, or gene expression. Technological advances in single-cell profiling have enabled researchers to assay multiple ... Full text Link to item Cite

Deciphering the species-level structure of topologically associating domains

Preprint · 2021 Summary Chromosome conformation capture technologies such as Hi-C have revealed a rich hierarchical structure of chromatin, with topologically associating domains (TADs) as a key organizational unit, but experimentally reported TAD architectures, ... Full text Cite

Proteomic and functional genomic landscape of receptor tyrosine kinase and ras to extracellular signal-regulated kinase signaling.

Journal article Sci Signal · October 25, 2011 Characterizing the extent and logic of signaling networks is essential to understanding specificity in such physiological and pathophysiological contexts as cell fate decisions and mechanisms of oncogenesis and resistance to chemotherapy. Cell-based RNA in ... Full text Link to item Cite

Sparse estimation for structural variability.

Journal article Algorithms Mol Biol · April 19, 2011 BACKGROUND: Proteins are dynamic molecules that exhibit a wide range of motions; often these conformational changes are important for protein function. Determining biologically relevant conformational changes, or true variability, efficiently is challengin ... Full text Link to item Cite

IsoBase: a database of functionally related proteins across PPI networks.

Journal article Nucleic Acids Res · January 2011 We describe IsoBase, a database identifying functionally related proteins, across five major eukaryotic model organisms: Saccharomyces cerevisiae, Drosophila melanogaster, Caenorhabditis elegans, Mus musculus and Homo Sapiens. Nearly all existing algorithm ... Full text Link to item Cite

Sparse estimation for structural variability

Conference Lecture Notes in Computer Science Including Subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics · November 10, 2010 Proteins are dynamic molecules that exhibit a wide range of motions; often these conformational changes are important for protein function. Determining biologically relevant conformational changes, or true variability, efficiently is challenging due to the ... Full text Cite

Struct2Net: a web service to predict protein-protein interactions using a structure-based approach.

Journal article Nucleic Acids Res · July 2010 Struct2Net is a web server for predicting interactions between arbitrary protein pairs using a structure-based approach. Prediction of protein-protein interactions (PPIs) is a central area of interest and successful prediction would provide leads for exper ... Full text Link to item Cite

IsoRankN: spectral methods for global alignment of multiple protein networks.

Conference Bioinformatics · June 15, 2009 MOTIVATION: With the increasing availability of large protein-protein interaction networks, the question of protein network alignment is becoming central to systems biology. Network alignment is further delineated into two sub-problems: local alignment, to ... Full text Link to item Cite

Graph algorithms for biological systems analysis

Conference Proceedings of the Annual ACM SIAM Symposium on Discrete Algorithms · December 1, 2008 The post-genomic era has witnessed an explosion in the quality, quantity and variety of biological data-sequence, structure, and networks. However, when building computational models on these data, some abstractions recur often. In particular, graph-based ... Cite

Global alignment of multiple protein interaction networks with application to functional orthology detection.

Journal article Proc Natl Acad Sci U S A · September 2, 2008 Protein-protein interactions (PPIs) and their networks play a central role in all biological processes. Akin to the complete sequencing of genomes and their comparative analysis, complete descriptions of interactomes and their comparative analysis is funda ... Full text Link to item Cite

Global alignment of multiple protein interaction networks.

Conference Pac Symp Biocomput · 2008 UNLABELLED: We describe an algorithm for global alignment of multiple protein-protein interaction (PPI) networks, the goal being to maximize the overall match across the input networks. The intuition behind our algorithm is that a protein in one PPI networ ... Link to item Cite

Predicting and annotating catalytic residues: an information theoretic approach.

Journal article J Comput Biol · October 2007 We introduce a computational method to predict and annotate the catalytic residues of a protein using only its sequence information, so that we describe both the residues' sequence locations (prediction) and their specific biochemical roles in the catalyze ... Full text Link to item Cite

Pairwise global alignment of protein interaction networks by matching neighborhood topology

Conference Lecture Notes in Computer Science Including Subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics · January 1, 2007 We describe an algorithm, ISORANK, for global alignment of two protein-protein interaction (PPI) networks. ISORANK aims to maximize the overall match between the two networks; in contrast, much of previous work has focused on the local alignment problem- i ... Full text Cite

Probabilistic modeling of systematic errors in two-hybrid experiments.

Conference Pac Symp Biocomput · 2007 UNLABELLED: We describe a novel probabilistic approach to estimating errors in two-hybrid (2H) experiments. Such experiments are frequently used to elucidate protein-protein interaction networks in a high-throughput fashion; however, a significant challeng ... Link to item Cite

Struct2net: integrating structure into protein-protein interaction prediction.

Conference Pac Symp Biocomput · 2006 UNLABELLED: This paper presents a framework for predicting protein-protein interactions (PPI) that integrates structure-based information with other functional annotations, e.g. GO, co-expression and co-localization, etc., Given two protein sequences, the ... Link to item Cite

Active learning for sampling in time-series experiments with application to gene expression analysis

Conference Icml 2005 Proceedings of the 22nd International Conference on Machine Learning · December 1, 2005 Many time-series experiments seek to estimate some signal as a continuous function of time. In this paper, we address the sampling problem for such experiments: determining which time-points ought to be sampled in order to minimize the cost of data collect ... Cite

Chaintweak: sampling from the neighbourhood of a protein conformation.

Conference Pac Symp Biocomput · 2005 When searching for an optimal protein structure, it is often necessary to generate a set of structures similar, e.g., within 4A Root Mean Square Deviation (RMSD), to some base structure. Current methods to do this are designed to produce only small deviati ... Link to item Cite

Identifying structural motifs in proteins.

Journal article Pac Symp Biocomput · 2003 In biological macromolecules, structural patterns (motifs) are often repeated across different molecules. Detection of these common motifs in a new molecule can provide useful clues to the functional properties of such a molecule. We formulate the problem ... Link to item Cite