Biology subjects

Mohanty, S.

Publications and source records attributed to Mohanty, S..

6 recordsLinked to original sources

bioRxiv2026

Towards Sparse Causal Features for Zero-shot Mutation Effect Prediction in a Protein Language Model

Protein language models (pLMs) such as ESM-2 achieve strong zero-shot mutation-effect prediction, yet the internal computations supporting these predictions remain poorly understood. We introduce a sparse feature circuit framework that combines sparse autoencoders, integrated-gradients attribution, and activation patching to identify the latent features that causally mediate zero-shot mutation effect prediction in ESM-2 650M. We evaluate this framework over 67 mutations ranging from strongly deleterious to weakly deleterious in the DNAJA1 J-domain, where ESM-2 predictions agree strongly with deep mutational scanning measurements. We find that circuits selected by indirect effect recover the model's predictions more efficiently and provide more informative biological explanations than those selected by raw activation changes, showing that activation magnitude does not necessarily reflect causal importance. We find that related substitutions reuse substantial portions of their recovered circuits, ranging from 40% to 75%, and that the shared features often represent residues in three-dimensional contact with the mutation site. To our knowledge, our work provides the first causal, feature-level account of zero-shot mutation effect prediction in a pLM.

bioinformatics↗

bioRxiv2018

Macrophage migration inhibitory factor (MIF) of Syrian golden hamster (Mesocricetus auratus) has similar structure and function as human MIF and promotes pancreatic tumor growth in vivo

Macrophage migration inhibitory factor (MIF) is a pleiotropic cytokine that increasingly is being studied in cancers and inflammatory diseases. Though murine models have been instrumental in understanding the functional role of MIF in different pathological conditions, the information obtained from these models is biased towards a specific species. In experimental science, results obtained from multiple clinically relevant animal models always provide convincing data that might recapitulate in humans. Syrian golden hamster (Mesocricetus auratus), is a clinically relevant animal model for multiple human diseases. Hence, the major objectives of this study were to characterize structure and function of hamster MIF, and finally evaluate its effect on pancreatic tumor growth in vivo. Initially, the recombinant hamster MIF (rha-MIF) was cloned, expressed and purified in bacterial expression system. The rha-MIF primary sequence, biochemical properties and crystal structure analysis showed a greater similarity with human MIF. The crystal structure of hamster MIF illustrates that it forms a homotrimer as known in human and mouse. However, hamster MIF exhibits some minor structural variations when compared to human and mouse MIF. The in vitro functional studies show that rha-MIF has tautomerase activity and enhances activation and migration of hamster peripheral blood mononuclear cells (PBMCs). Interestingly, injection of rha-MIF into HapT1 pancreatic tumor bearing hamsters significantly enhanced the tumor growth and tumor associated angiogenesis. Together, the current study shows a structural and functional similarity between hamster and human MIF. Moreover, it has demonstrated that a high-level of circulating MIF originating from non-tumor cells might also promote pancreatic tumor growth in vivo.

biochemistry↗

bioRxiv2018

Taking a Dive: Experiments in Deep Learning for Automatic Ontology-based Annotation of Scientific Literature

I.Text mining approaches for automated ontology-based curation of biological and biomedical literature have largely focused on syntactic and lexical analysis along with machine learning. Recent advances in deep learning have shown increased accuracy for textual data annotation. However, the application of deep learning for ontology-based curation is a relatively new area and prior work has focused on a limited set of models.\n\nHere, we introduce a new deep learning model/architecture based on combining multiple Gated Recurrent Units (GRU) with a character+word based input. We use data from five ontologies in the CRAFT corpus as a Gold Standard to evaluate our models performance. We also compare our model to seven models from prior work. We use four metrics - Precision, Recall, F1 score, and a semantic similarity metric (Jaccard similarity) to compare our models output to the Gold Standard. Our model resulted in a 84% Precision, 84% Recall, 83% F1, and a 84% Jaccard similarity. Results show that our GRU-based model outperforms prior models across all five ontologies. We also observed that character+word inputs result in a higher performance across models as compared to word only inputs.\n\nThese findings indicate that deep learning algorithms are a promising avenue to be explored for automated ontology-based curation of data. This study also serves as a formal comparison and guideline for building and selecting deep learning models and architectures for ontology-based curation.

bioinformatics↗

bioRxiv2018

An Atlas of Human and Murine Genetic Influences on Osteoporosis

Osteoporosis is a common debilitating chronic disease diagnosed primarily using bone mineral density (BMD). We undertook a comprehensive assessment of human genetic determinants of bone density in 426,824 individuals, identifying a total of 518 genome-wide significant loci, (301 novel), explaining 20% of the total variance in BMD--as estimated by heel quantitative ultrasound (eBMD). Next, meta-analysis identified 13 bone fracture loci in ~1.2M individuals, which were also associated with BMD. We then identified target genes from cell-specific genomic landscape features, including chromatin conformation and accessible chromatin sites, that were strongly enriched for genes known to influence bone density and strength (maximum odds ratio = 58, P = 10-75). We next performed rapid throughput skeletal phenotyping of 126 knockout mice lacking eBMD Target Genes and showed that these mice had an increased frequency of abnormal skeletal phenotypes compared to 526 unselected lines (P < 0.0001). In-depth analysis of one such Target Gene, DAAM2, showed a disproportionate decrease in bone strength relative to mineralization. This comprehensive human and murine genetic atlas provides empirical evidence testing how to link associated SNPs to causal genes, offers new insights into osteoporosis pathophysiology and highlights opportunities for drug development.

genomics↗

bioRxiv2018

22-hydroxycholesterol leads to efficient dopaminergic specification of human Mesenchymal Stem Cells

In the current study, we aim to investigate the neurogenic effect of 22 (R)- hydroxycholesterol (22-HC), on hMSCs obtained from bone marrow, adipose tissue and dental pulp. The effect was evaluated on the basis of detailed morphological and morphometric analysis, expression of genes and proteins associated with maturation of neurons (NF, MAP2, TUJ1, TH), channel ion proteins (Kv4.2 & SCN5A), chemical functionality (DAT), synapse forming tendency of neurons (Synaptophysin) and transcription factors (ngn2 & Pitx3), efficiency of generation of dopaminergic neurons and functional assessment. The percentage of non- DA cells generation (Ach, TPH2, S100 and GFAP positive cells) was also evaluated to confirm selective neurogenic potential of 22-HC. Post analysis, it was observed that 22-HC yields higher percentage of functional DA neurons and has differential effect on various tissue- specific primary human MSCs. This study, which is one of its kinds, may help in improvising the approach of cell based treatment regimes and drug testing in pharmaceutical industry for Parkinsons disease.

neuroscience↗

bioRxiv2017

A 23bp Indel Polymorphism in TLR2 Gene Enhances Inflammation and Disease Severity in Dengue

BackgroundDengue is the most rapidly spreading viral disease transmitted by the bite of infected Aedes mosquitos. Pathogenesis of dengue is still unclear; although host genetic factors, immune responses and virus serotypes have been proposed to contribute to disease severity. The development of high-throughput methods have allowed to scale up capabilities of identifying the key markers of inflammation. Since NS1 protein of dengue virus has been reported to activate immune cells towards enhanced inflammation through TLR2, we examined the role of a polymorphism, a 23bp deletion in 5UTR region of TLR2 gene in patients with dengue (with and without warning signs) and correlated with plasma levels of inflammatory mediators with disease severity and viral serotypes.\n\nMethodsEighty nine patients classified as per WHO 2009 criteria during dengue outbreak in Odisha, India in 2016 were included in the current study. Presence of dengue virus (DENV) was demonstrated by detecting NS1 antigen, IgM capture ELISA and serotypes in circulation were discriminated by type-specific RT-PCR and/or sequencing. Sixty-one confirmed dengue cases were typed for TLR2 indel polymorphism and compared with 485 disease free controls. Plasma samples were assayed for 41-plex cytokine/ chemokines using Luminex bead based immunoassay.\n\nResultsPresence of 23bp deletion allele of TLR2 gene was significantly more in patients with severe dengue in comparison to dengue fever cases (p= 0.03; Odds ratio 4.05) although the frequency of insertion (Ins) allele of TLR2 was comparable in healthy controls and dengue cases (82.4 and 87.9 % respectively). Seventy-three (82%) samples were found to be positive by NS1/IgM capture ELISA/ RT-PCR. DENV-2 was predominant (58%) during the outbreak. Among the host inflammatory biomarkers 9 molecules were significantly altered in dengue patients when compared to healthy controls. The increased levels of IFN-{gamma}, GM-CSF, IL-10, IL-1R and MIP-1{beta} correlated significantly with severe dengue.\n\nConclusionsThe frequency of 23bp Indel mutation of TLR2 was comparable between healthy controls and dengue fever (with and without warning signs), suggesting that this indel mutation does not contribute significantly to susceptibility/ resistance to dengue; however, del allele of TLR2 gene was significantly more associated in patients with severe dengue symptoms when compared to dengue fever cases.

microbiology↗