Biology subjects

Ullanat, V.

Publications and source records attributed to Ullanat, V..

4 recordsLinked to original sources

bioRxiv2025

Learning the language of protein-protein interactions

Protein Language Models (PLMs) trained on large databases of protein sequences have proven effective in modeling protein biology across a wide range of applications. However, while PLMs excel at capturing individual protein properties, they face challenges in natively representing protein-protein interactions (PPIs), which are crucial to understanding cellular processes and disease mechanisms. Here, we introduce MINT, a PLM specifically designed to model sets of interacting proteins in a contextual and scalable manner. Using unsupervised training on a large curated PPI dataset derived from the STRING database, MINT outperforms existing PLMs in diverse tasks relating to protein-protein interactions, including binding affinity prediction and estimation of mutational effects. Beyond these core capabilities, it excels at modeling interactions in complex protein assemblies and surpasses specialized models in antibody-antigen modeling and T cell receptor-epitope binding prediction. MINTs predictions of mutational impacts on oncogenic PPIs align with experimental studies, and it provides reliable estimates for the potential for cross-neutralization of antibodies against SARS-CoV-2 variants of concern. These findings position MINT as a powerful tool for elucidating complex protein interactions, with significant implications for biomedical research and therapeutic discovery.

bioinformatics↗

bioRxiv2024

A deep learning method for predicting interactions for intrinsically disordered regions of proteins

Intrinsically disordered proteins or regions (IDPs/IDRs) adopt diverse binding modes with different partners, from coupled-folding-and-binding, to fuzzy binding, to fully-disordered binding. Characterizing IDR interfaces is challenging experimentally and computationally. The state-of-the-art AlphaFold-multimer and AlphaFold3 can be used to predict IDR binding sites, although they are less accurate at their benchmarked confidence cutoffs. Here, we developed Disobind, a deep-learning method that predicts inter-protein contact maps and interface residues for an IDR and its partner, given their sequences. It uses sequence embeddings from the ProtT5 protein language model. Disobind outperforms state-of-the-art interface predictors for IDRs. It also outperforms AlphaFold-multimer and AlphaFold3 at multiple confidence cutoffs. Combining Disobind and AlphaFold-multimer predictions further improves the performance. In contrast to current methods, Disobind considers the context of the binding partner and does not depend on structures and multiple sequence alignments. Its predictions can be used to localize IDRs in large assemblies and characterize IDR-mediated interactions. A record of this papers transparent peer review process is included in the supplemental information.

bioinformatics↗

bioRxiv2021

Landscape of epithelial-mesenchymal plasticity as an emergent property of coordinated teams in regulatory networks

Elucidating the design principles of regulatory networks driving cellular decision-making has fundamental implications in mapping and eventually controlling cell-fate decisions. Despite being complex, these regulatory networks often only give rise to a few phenotypes. Previously, we identified two "teams" of nodes in a small cell lung cancer regulatory network that constrained the phenotypic repertoire and aligned strongly with the dominant phenotypes obtained from network simulations (Chauhan et al., 2021). However, it remained elusive whether these "teams" exist in other networks, and how do they shape the phenotypic landscape. Here, we demonstrate that five different networks of varying sizes governing epithelial-mesenchymal plasticity comprised of two "teams" of players - one comprised of canonical drivers of epithelial phenotype and the other containing the mesenchymal inducers. These "teams" are specific to the topology of these regulatory networks and orchestrate a bimodal phenotypic landscape with the epithelial and mesenchymal phenotypes being more frequent and dynamically robust to perturbations, relative to the intermediary/hybrid epithelial/ mesenchymal ones. Our analysis reveals that network topology alone can contain information about corresponding phenotypic distributions, thus obviating the need to simulate them. We propose "teams" of nodes as a network design principle that can drive cell-fate canalization in diverse decision-making processes.

systems biology↗

bioRxiv2021

Emergent properties of coupled bistable switches

Understanding the dynamical hallmarks of network motifs is one of the fundamental aspects of systems biology. Positive feedback loops constituting one or two nodes - self-activation, toggle switch, and double activation loops - are commonly observed motifs in regulatory networks underlying cell-fate decision systems. Their individual dynamics are well-studied; they are capable of exhibiting bistability. However, studies across various biological systems suggest that such positive feedback loops are interconnected with one another, and design principles of coupled bistable motifs remain unclear. We wanted to ask what happens to bistability or multistability traits and the phenotypic space (collection of phenotypes exhibited by a system) due to the couplings. In this study, we explore a set of such interactions using discrete and continuous simulation methods. Our results suggest that couplings that do not connect the bistable switches in a way that contradicts the connections within individual bistable switches lead to a steady state space that is strictly a subset of the set of possible combinations of steady states of bistable switches. Furthermore, adding direct and indirect self-activations to these coupled networks can increase the frequency of multistability. Thus, our observations reveal specific dynamical traits exhibited by various coupled bistable motifs.

systems biology↗