Meghana Kshirsagar

Meghana Kshirsagar

Auto-generated profile, based on publicly available information

Meghana Kshirsagar

AI for Good Research Lab, Microsoft Corporation, Redmond, USA

Research Focus & Metrics

Academic Career Length

10Years (2017-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
27 February 2025 in Nature Communications
Verified
Journal Article
Open access
PDF file
MKMeghana Kshirsagar
Meghana Kshirsagar
AMArtur Meller
Artur Meller
IHIan R. Humphreys
Ian R. Humphreys
SSSamuel Sledzieski
Samuel Sledzieski
YXYixi Xu
Yixi Xu
MBMinkyung Baek
Minkyung Baek
The majority of proteins must form higher-order assemblies to perform their biological functions, yet few machine learning models can accurately and rapidly predict the symmetry of assemblies involving multiple copies of the same protein chain. Here, we address this gap by finetuning several classes of protein foundation models, to predict homo-oligomer symmetry. Our best model named Seq2Symm, which utilizes ESM2, outperforms existing template-based and deep learning methods achieving an average AUC-PR of 0.47, 0.44 and 0.49 across homo-oligomer symmetries on three held-out test sets compared to 0.24, 0.24 and 0.25 with template-based search. Seq2Symm uses a single sequence as input and can predict at the rate of ~80,000 proteins/hour. We apply this method to 5 proteomes and ~3.5 million unlabeled protein sequences, showing its promise to be used in conjunction with downstream computationally intensive all-atom structure generation methods such as RoseTTAFold2 and AlphaFold2-multimer. Code, datasets, model are available at: https://github.com/microsoft/seq2symm.

15 August 2022 in Genome Biology
Verified
Journal Article
Open access
PDF file
MKMeghana Kshirsagar
Meghana Kshirsagar
HYHan Yuan
Han Yuan
JFJuan Lavista Ferres
Juan Lavista Ferres
CLChristina Leslie
Christina Leslie
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. On the task of retrieving the motifs of expressed TFs in a given cell type, BindVAE is competitive with existing motif discovery approaches.

10 November 2023 in bioRxiv
Verified
Preprint
Open access
PDF file
SSSamuel Sledzieski
Samuel Sledzieski
MKMeghana Kshirsagar
Meghana Kshirsagar
MBMinkyung Baek
Minkyung Baek
BBBonnie Berger
Bonnie Berger
RDRahul Dodhia
Rahul Dodhia
JFJuan Lavista Ferres
Juan Lavista Ferres
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.

10 January 2025 in bioRxiv
Verified
Preprint
Open access
PDF file
PPPreethi K. Periyakoil
Preethi K. Periyakoil
MSMelanie H. Smith
Melanie H. Smith
MKMeghana Kshirsagar
Meghana Kshirsagar
DRDaniel Ramirez
Daniel Ramirez
EDEdward F. DiCarlo
Edward F. DiCarlo
CLChristina S. Leslie
Christina S. Leslie
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.

29 December 2021 in GigaScience
Verified
Journal Article
Open access
PDF file
JLJeffrey N Law
Jeffrey N Law
KAKyle Akers
Kyle Akers
NTNure Tasnina
Nure Tasnina
CSCatherine M Della Santina
Catherine M Della Santina
SDShay Deutsch
Shay Deutsch
TMT M Murali
T M Murali
Background: Network propagation has been widely used for nearly 20 years to predict gene functions and phenotypes. Despite the popularity of this approach, little attention has been paid to the question of provenance tracing in this context, e.g., determining how much any experimental observation in the input contributes to the score of every prediction. Results: We design a network propagation framework with 2 novel components and apply it to predict human proteins that directly or indirectly interact with SARS-CoV-2 proteins. First, we trace the provenance of each prediction to its experimentally validated sources, which in our case are human proteins experimentally determined to interact with viral proteins. Second, we design a technique that helps to reduce the manual adjustment of parameters by users. We find that for every top-ranking prediction, the highest contribution to its score arises from a direct neighbor in a human protein-protein interaction network. We further analyze these results to develop functional insights on SARS-CoV-2 that expand on known biology such as the connection between endoplasmic reticulum stress, HSPA5, and anti-clotting agents. Conclusions: We examine how our provenance-tracing method can be generalized to a broad class of network-based algorithms. We provide a useful resource for the SARS-CoV-2 community that implicates many previously undocumented proteins with putative functional relationships to viral infection. This resource includes potential drugs that can be opportunistically repositioned to target these proteins. We also discuss how our overall framework can be extended to other, newly emerging viruses.

29 June 2022 in bioRxiv
Verified
Preprint
Open access
PDF file
AMArtur Meller
Artur Meller
MWMichael Ward
Michael Ward
JBJonathan Haddad Borowsky
Jonathan Haddad Borowsky
JLJeffrey M. Lotthammer
Jeffrey M. Lotthammer
MKMeghana Kshirsagar
Meghana Kshirsagar
GBGregory R. Bowman
Gregory R. Bowman
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.