Biology subjects

Kshirsagar, M.

Publications and source records attributed to Kshirsagar, M..

7 recordsLinked to original sources

bioRxiv2025

Accelerating Biomolecular Modeling with AtomWorks andRF3

Deep learning methods trained on protein structure databases have revolutionized biomolecular structure prediction, but developing and training new models remains a considerable challenge. To facilitate the development of new models, we present AtomWorks: a broadly applicable data framework for developing state-of-the-art biomolecular foundation models spanning diverse tasks, including structure prediction, generative protein design, and fixed backbone sequence design. We use AtomWorks to train RosettaFold-3 (RF3), a structure prediction network capable of predicting arbitrary biomolecular complexes with an improved treatment of chirality that narrows the performance gap between closed-source AlphaFold3 (AF3) and existing open-source implementations. We expect that AtomWorks will accelerate the next generation of open-source biomolecular machine learning models and that RF3 will be broadly useful as a structure prediction tool. To this end, we release the AtomWorks framework (https://github.com/RosettaCommons/atomworks), together with curated training data, code and model weights for RF3 (https://github.com/RosettaCommons/modelforge) under a permissive BSD license.

biochemistry↗

bioRxiv2025

Deep topic modeling of spatial transcriptomics in the rheumatoid arthritis synovium identifies distinct classes of ectopic lymphoid structures

Single-cell RNA sequencing studies have revealed the heterogeneity of cell states present in the rheumatoid arthritis (RA) synovium. However, it remains unclear how these cell types interact with one another in situ and how synovial microenvironments shape observed cell states. Here, we use spatial transcriptomics (ST) to define stable microenvironments across eight synovial tissue samples from six RA patients and characterize the cellular composition of ectopic lymphoid structures (ELS). To identify disease-relevant cellular communities, we developed DeepTopics, a scalable reference-free deconvolution method based on a Dirichlet variational autoencoder architecture. DeepTopics identified 22 topics across tissue samples that were defined by specific cell types, activation states, and/or biological processes. Some topics were defined by multiple colocalizing cell types, such as CD34+ fibroblasts and LYVE1+ macrophages, suggesting functional interactions. Within ELS, we discovered two divergent cellular patterns that were stable across ELS in each patient and typified by the presence or absence of a "germinal-center-like" topic. DeepTopics is a versatile and computationally efficient method for identifying disease-relevant microenvironments from ST data, and our results highlight divergent cellular architectures in histologically similar RA synovial samples that have implications for disease pathogenesis.

immunology↗

bioRxiv2023

Democratizing Protein Language Models with Parameter-Efficient Fine-Tuning

Proteomics has been revolutionized by large pre-trained protein language models, which learn unsupervised representations from large corpora of sequences. The parameters of these models are then fine-tuned in a supervised setting to tailor the model to a specific downstream task. However, as model size increases, the computational and memory footprint of fine-tuning becomes a barrier for many research groups. In the field of natural language processing, which has seen a similar explosion in the size of models, these challenges have been addressed by methods for parameter-efficient fine-tuning (PEFT). In this work, we newly bring parameter-efficient fine-tuning methods to proteomics. Using the parameter-efficient method LoRA, we train new models for two important proteomic tasks: predicting protein-protein interactions (PPI) and predicting the symmetry of homooligomers. We show that for homooligomer symmetry prediction, these approaches achieve performance competitive with traditional fine-tuning while requiring reduced memory and using three orders of magnitude fewer parameters. On the PPI prediction task, we surprisingly find that PEFT models actually outperform traditional fine-tuning while using two orders of magnitude fewer parameters. Here, we go even further to show that freezing the parameters of the language model and training only a classification head also outperforms fine-tuning, using five orders of magnitude fewer parameters, and that both of these models outperform state-of-the-art PPI prediction methods with substantially reduced compute. We also demonstrate that PEFT is robust to variations in training hyper-parameters, and elucidate where best practices for PEFT in proteomics differ from in natural language processing. Thus, we provide a blueprint to democratize the power of protein language model tuning to groups which have limited computational resources.

bioinformatics↗

bioRxiv2022

Predicting the locations of cryptic pockets from single protein structures using the PocketMiner graph neural network

Cryptic pockets expand the scope of drug discovery by enabling targeting of proteins currently considered undruggable because they lack pockets in their ground state structures. However, identifying cryptic pockets is labor-intensive and slow. The ability to accurately and rapidly predict if and where cryptic pockets are likely to form from a protein structure would greatly accelerate the search for druggable pockets. Here, we present PocketMiner, a graph neural network trained to predict where pockets are likely to open in molecular dynamics simulations. Applying PocketMiner to single structures from a newly-curated dataset of 39 experimentally-confirmed cryptic pockets demonstrates that it accurately identifies cryptic pockets (ROC-AUC: 0.87) >1,000-fold faster than existing methods. We apply PocketMiner across the human proteome and show that predicted pockets open in simulations, suggesting that over half of proteins thought to lack pockets based on available structures are likely to contain cryptic pockets, vastly expanding the druggable proteome.

biophysics↗

bioRxiv2022

An epigenetic barrier sets the timing of human neuronal maturation

The pace of human brain development is highly protracted compared to most other species. The maturation of cortical neurons is particularly slow, extending over months to years to reach adult-like functions. Remarkably, such protracted timing is retained in cortical neurons derived from human pluripotent stem cell (hPSC) during in vitro differentiation or upon transplantation into the murine brain in vivo. Those findings suggest the presence of a cell intrinsic clock that sets the pace of neuronal maturation, though the molecular nature of such a clock has remained elusive. Here, we identify an epigenetic developmental program which sets the timing of human neuronal maturation. First, we developed a human PSC-based approach to synchronize the birth of cortical neurons in vitro which allowed us to define a detailed atlas of progressive morphological, functional, and molecular maturation in human cortical neurons. Interestingly, we observed a slow, temporal unfolding of maturation programs that is limited by the retention of a specific set of epigenetic factors. Loss-of-function studies for several of those factors in cortical neurons enables precocious molecular and physiological maturation. Remarkably, transient inhibition of EZH2, EHMT1/2 or DOT1L, at the progenitor stage primes newly born neurons to rapidly acquire mature properties upon differentiation. Therefore, our findings reveal that the rate at which human neurons mature is set well before neurogenesis through the establishment of an "epigenetic barrier" in progenitor cells. Mechanistically, this barrier acts by holding transcriptional maturation programs in a poised state that gets gradually released during neuronal differentiation to ensure the prolonged timeline characteristic of human cortical neuron maturation.

developmental biology↗

bioRxiv2021

Dirichlet variational autoencoders for de novo motif discovery from accessible chromatin

We present a novel unsupervised deep learning approach called BindVAE, based on Dirichlet variational autoencoders, for jointly decoding multiple TF binding signals from open chromatin regions. BindVAE can disentangle an input DNA sequence into distinct latent factors that encode cell-type specific in vivo binding signals for individual TFs, composite patterns for TFs involved in cooperative binding, and genomic context surrounding the binding sites. For the task of retrieving motifs of expressed TFs for a given cell type, we find that BindVAE has a higher precision, albeit lower recall, compared to other motif discovery approaches.

bioinformatics↗

bioRxiv2018

BindSpace: decoding transcription factor binding signals by large-scale joint embedding

Decoding transcription factor (TF) binding signals in genomic DNA is a fundamental problem. Here we present a prediction model called BindSpace that learns to embed DNA sequences and TF class/family labels into the same space. By training on binding data for hundreds of TFs and embedding over 1M DNA sequences, BindSpace achieves state-of-the-art multiclass binding prediction performance, in vitro and in vivo, and can distinguish signals of closely related TFs.

bioinformatics↗