Biology subjects

Sahni, H.

Publications and source records attributed to Sahni, H..

3 recordsLinked to original sources

bioRxiv2026

Polynomial Trajectory Compression for Protein Language Model Embeddings

Protein language models (PLMs) generate rich, layer-wise embeddings that capture diverse biological information but are expensive in terms of storage and computation at scale. In this work, we propose a compact surrogate representation for PLM embeddings across transformer layers using low-dimensional PCA projections and cubic polynomial trajectories. This approach enables efficient storage and on-demand reconstruction of these protein-level embeddings at any layer without rerunning the PLM. We evaluate our method on two downstream tasks: protein-protein interaction and subcellular localization using ESM-35M and ESM-3B PLM. We show that the surrogate embeddings achieve high reconstruction fidelity while reducing storage and computational requirements significantly. The new approach also retains downstream task prediction performance compared to original embeddings. Our approach provides a scalable and practical solution for large-scale protein embedding storage and reuse.

bioinformatics↗

bioRxiv2025

Challenges in predicting protein-protein interactions of understudied viruses: Arenavirus-Human interactions

Understanding protein-protein interactions (PPIs) between viruses and human proteins is crucial for uncovering infection mechanisms and identifying potential therapeutic targets. The ability to generalize PPI predictive models across understudied viruses presents a significant challenge. In this work, we use arenavirus-human PPIs to illustrate the difficulties associated with model generalization, which are compounded by a lack of both positive and negative data. We employ a Transfer Learning approach to investigate arenavirus-human PPI by utilizing models trained on better-studied virus-human and human-human interactions. Additionally, we curate and assess four types of negative sampling datasets to evaluate their impact on model performance. Despite the overall high accuracies (93-99%) and AUPRC scores (0.8-0.9) appearing promising, further analysis indicates that these performance metrics can be misleading due to data leakage, data bias, and overfitting, especially concerning under-represented viral proteins. We reveal these gaps and assess the impact of data imbalance through standard k-fold cross-validation and Independent Blind Testing with a Balanced Dataset, leading to a drop in accuracy below 50%. We propose a viral protein-specific evaluation framework that groups viral proteins into majority and minority classes based on their representation in the dataset, allowing for comparison of model performance across these groups using balanced accuracies. This framework offers a more robust evaluation of model generalizability, addressing biases inherent in standard evaluation techniques and paving the way for more reliable PPI prediction models for understudied viruses.

bioinformatics↗

bioRxiv2024

Insertion and Anchoring of HIV-1 Fusion Peptide into Complex Membrane Mimicking Human T-cell

A fundamental understanding of how HIV-1 envelope (Env) protein facilitates fusion is still lacking. The HIV-1 fusion peptide, consisting of 15 to 22 residues, is the N-terminus of the gp41 subunit of the Env protein. Further, this peptide, a promising vaccine candidate, initiates viral entry into target cells by inserting and anchoring into human immune cells. The influence of membrane lipid reorganization and the conformational changes of the fusion peptide during the membrane insertion and anchoring processes, which can significantly affect HIV-1 cell entry, remains largely unexplored due to the limitations of experimental measurements. In this work, we investigate the insertion of the fusion peptide into an immune cell membrane mimic through multiscale molecular dynamics simulations. We mimic the native T-cell by constructing a 9-lipid asymmetric membrane, along with geometrical restraints accounting for insertion in the context of gp41. To account for the slow timescale of lipid mixing while enabling conformational changes, we implement a protocol to go back and forth between atomistic and coarse-grained simulations. Our study provides a molecular understanding of the interactions between the HIV-1 fusion peptide and the T-cell membrane, highlighting the importance of conformational flexibility of fusion peptides and local lipid reorganization in stabilizing the anchoring of gp41 into the targeted host membrane during the early events of HIV-1 cell entry. Importantly, we identify a motif within the fusion peptide critical for fusion that can be further manipulated in future immunological studies. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=127 SRC="FIGDIR/small/606381v1_ufig1.gif" ALT="Figure 1"> View larger version (50K): org.highwire.dtl.DTLVardef@d80eedorg.highwire.dtl.DTLVardef@bc2482org.highwire.dtl.DTLVardef@f715org.highwire.dtl.DTLVardef@15dba11_HPS_FORMAT_FIGEXP M_FIG Table of Content. C_FIG

biophysics↗