Biology subjects

Joshi, H.

Publications and source records attributed to Joshi, H..

At least 19 recordsLinked to original sources

bioRxiv2026

BiofilmQ-HT: An integrated browser-based workflow for standardized analysis of high-throughput crystal violet-based biofilm assays

The crystal violet (CV) microtiter plate assay is widely used to compare microbial biofilm formation because it is inexpensive, simple, and readily adapted to 96-well formats. Its high-throughput design enables multiple strains, growth conditions, and treatments to be tested in parallel, but increasing experimental complexity can make downstream data handling difficult. Plate layouts, absorbance measurements, blank correction, replicate analysis, and visualization are often managed using separate spreadsheets or software, requiring repeated transfer and organization of experimental information. This can compromise consistency in linking measurements with their corresponding conditions and controls. We developed BiofilmQ-HT, a standalone, browser-based workflow that integrates plate annotation with direct import of plate-reader data, condition-specific blank correction, replicate summarization, growth-normalized specific biofilm formation (SBF) analysis, percentage-inhibition analysis, visualization, and data export. The workflow was applied to three representative datasets: a comparison of biofilm formation in 25% and 100% media, an analysis of bacteriophage- and antibiotic-associated antibiofilm activity, and a screening of biofilm formation among natural microbial isolates. These applications demonstrate a common analytical framework for comparative biofilm analysis, antibiofilm screening, and high-throughput isolate screening while retaining experimental conditions and controls. By linking plate organization to quantitative analysis and reporting, BiofilmQ-HT provides a reproducible and traceable approach for processing condition-rich CV assay datasets and facilitates consistent comparisons across experimental groups. The utility therefore improves consistency in data handling without altering the underlying experimental assay or requiring changes to the established assay.

microbiology↗

bioRxiv2026

SketchDNA: A GUI-Enabled Toolkit for Multiscale Modeling of Topological DNA Structures

Computational modeling tools have enabled detailed exploration of the structural dynamics of nucleic acids at the nanoscale. Despite these developments, a unified platform for creating multiscale models of topological DNA structures, which are of fundamental importance across rapidly converging biological disciplines, is lacking. Here we present a GUI-enabled topological DNA design tool, SketchDNA (SDNA), that facilitates a user-friendly interface to create all-atom (AA) and coarse-grained (CG) models of DNA structures with tunable topological parameters. Building on the modular and open-source framework of SDNA, the software can be readily integrated with both emerging and existing DNA design platforms, such as oxDNA, MrDNA, etc. We demonstrate the utility of SDNA computational framework by simulating the conformational dynamics of three representative topological DNA structures, minicircles, catenanes, and Borromean rings. Analyzing AA and CG molecular dynamics (MD) simulations of these topological DNA systems using the AMBER and Martini DNA force fields, respectively, we characterize their equilibrium structural dynamics, fluctuations, and topological properties. While the AA simulations allow us to characterize the topology-dependent structural rearrangements at the nucleotide level, long-timescale CG simulations reveal supercoiling-induced conformational transitions. The multiscale SDNA toolkit broadens the applications of MD simulations by enabling in-situ characterization of the biophysical properties of topological DNA nanostructures. The GUI version of SDNA is available at https://sdna.biotech.iith.ac.in without registration, while the source code is accessible at GitHub.

biophysics↗

bioRxiv2026

Mechanistic Insights into Unfolding of the Mitochondrial Localization Sequence M1 of TDP-43 and Identification of a MAPKAPK2 Inhibitor as a potential binder of M1 for inhibition of TDP-43's Aberrant Mitochondrial Localization

Mechanistic elucidation and inhibition of the pathogenic aberrant mitochondrial localization of the RNA/DNA-binding protein, TDP-43, can help in the therapeutics of the neurodegenerative disease amyotrophic lateral sclerosis (ALS). A mitochondrial localization sequence of TDP-43, M1, is largely solvent inaccessible, therefore, how it interacts with the mitochondrial import machinery to facilitate TDP-43s transit to mitochondria is unclear. Towards this, we examined the unfolding TDP-43s N-terminal domain (NTD) that hosts M1, using equilibrium all-atom molecular dynamics (MD) simulations, and observed an early loss of the hydrogen-bonded interactions between {beta}4-{beta}5 bridge and the interactions involving residues Phe-35 and Gly-40 of M1, indicating structural lability of M1 to become solvent-accessible that may enhance its interaction with the mitochondrial receptor(s) for import. Furthermore, via virtual screening of 2,115 FDA-approved and 515,545 non-FDA-approved small molecules from ZINC15 database towards binding to M1 and inhibiting TDP-43s mitochondrial import, we identified a molecule, ZINC73240059, that was previously characterized as an inhibitor of MAP kinase-activating protein kinase 2 (MAPKAPK2). ZINC73240059 remains stably bound to M1 of NTD during MD simulations manifesting negative Gibbs free energy ({Delta}G) with significant contribution from Pro-36 of M1. Overall, ZINC73240059 can be a molecule of interest towards thwarting TDP-43s pathogenic mitochondrial localization in ALS.

biophysics↗

bioRxiv2026

Residue-specific dominant-negative mutant of ubiquitin reveals functional selectivity as Ubp14 deubiquitinase inhibitor

Multifunctional proteins encode specificity through nuanced molecular interactions that operate in the presence of abundant wild-type molecules. Previous studies on multifunctional proteins have shown how certain regulated interactions parse complex biological activities into discrete functional modules. Discrete interaction edges can function as regulatory nodes whose perturbations selectively remodel proteostasis output, especially under disease conditions. However, whether such interaction specificity can be harnessed to selectively manipulate functional modules and reveal residue-level control principles, remains largely unexplored. Here, we screened for dominant negative mutations that impact partner-selective functions linked to Leu8, a critical binding residue on ubiquitin. Our results reveal that the Leu8Ala mutation specifically leads to accumulation of polyubiquitinated proteins and decreased levels of free ubiquitin suggesting loss of deubiquitinase function in yeast cells. Cellular and biochemical analyses establish that the Leu8Ala variant of ubiquitin specifically inhibits the yeast deubiquitinase, Ubp14, and its human homolog, USP5. We further demonstrate remodelling of the binding interface with increased interface contacts for the variant and Ubp14 complex. The variant shows a higher inhibitory potency compared to the wild-type ubiquitin and can inhibit Ubp14 both as unconjugated and as conjugated ubiquitin chain indicating the strength of its inhibitory function. Our results provide mechanistic insight into how edge perturbations in ubiquitin can reveal critical nodes that impact selective functions and thus fine-tune the cellular proteostasis network for therapeutic benefit.

biochemistry↗

bioRxiv2025

Elucidating the conformational dynamics of the mitochondrial localization signal, M3, of TDP-43 and accessing potential inhibitors using molecular docking and simulation

Aberrant mitochondrial localization of the RNA/DNA-binding protein TDP-43 is implicated in amyotrophic lateral sclerosis (ALS) which may affect mitochondrial dynamics and contribute to neuronal toxicity. Inhibitors of the cytoplasmic aggregation of TDP-43 were reported previously, but their effect on the mitochondrial mis-localization of TDP-43 and the mitochondrial function remains to be investigated. Three internal peptide sequences from TDP-43, M1, M3, and M5, were found to enable TDP-43s mitochondrial localization and competitive inhibition using these peptides thwarted mitochondrial import of TDP-43 and rescued TDP-43-induced cytotoxicity to neurons. Here, we performed a virtual in silico screening of 2,115 FDA-approved small molecules for affinity against the M3 region of TDP-43 (aa: 146 - 150) that lies in the RNA recognition motif-1 (RRM1) domain. Multiple all-atom MD simulations were carried out starting with two different conformations of the tandem RRMs to understand the dynamics of the target M3 region in explicit solvent water. The analysis of the simulation trajectories suggests that the M3 region is relatively non-flexible and buried relative to the other regions of the tandem RRM1-2 domains. Cholecalciferol (Vitamin D3), from virtual screening, docked consistently with the M3 region in different docking strategies even amidst the regions poor solvent accessibility. Vitamin D3 also remained stably bound to the M3 region in most frames of four replica MD simulations, each of one microsecond. Taken together, our study proposes vitamin D3 as a potential binder to the M3 region which may inhibit the pathogenic mitochondrial mis-localization of TDP-43.

bioinformatics↗

bioRxiv2025

Small amphiphilic DNA for programmable transmembrane signaling and amplification

Interrogating the sequence-specific DNA/RNA content of a living cell without destroying it, is an elusive goal in biology and diagnostics with transformative potential. A DNA/RNA-based approach, however, must relay the recognition event across the intact lipid bilayer, a step that has long been considered impossible because hydrophilic, massively charged DNA/RNA is incompatible with the hydrophobic membrane interior. Here, we show that DNA branch migration can be driven across a lipid bilayer, converting an intracellular nucleic-acid sequence into an amplified, extracellular optical signal while preserving cell viability and compartmentalization, a synthetic transmembrane signal transduction analogous to natural receptors, such as G-protein-coupled receptors (GPCRs), but operating in reverse. We demonstrate this principle with Hybridization Across Lipid for Oligonucleotide Sensing (HALOS), an amphiphilic DNA hairpin comprising a toehold for recognition, a stem for stability, a loop, and cholesterols for transmembrane anchoring. Upon binding to a sequence-specific nucleic acid target, toehold-mediated strand invasion drives DNA branch migration across the bilayer, switching HALOS from a closed to an open conformation that relays the signal to the opposite face. Molecular dynamics simulations revealed that a cholesterol belt stabilizes the DNA stem within the membrane, preserving the hairpin structure necessary for transmembrane signaling. Notably, this branch migration proceeds despite the kinetic barrier of hybridization through the hydrophobic bilayer interior. By combining HALOS with an isothermal hybridization chain reaction (HCR), we established a platform that enables intracellular nucleic acid target detection and amplified fluorescent reporting from outside synthetic vesicles and live mammalian cells, achieving nanomolar-range sensitivity. These results establish DNA branch migration across lipid membranes as a design principle for programmable lysis-free molecular sensing in intact cells.

biophysics↗

bioRxiv2025

Generating diffusion MRI scalar maps from T1-weighted images using Reversible GANs

Diffusion tensor imaging (DTI) provides valuable insights into brain tissue microstructure, but acquiring high-quality DTI data is time-intensive and not always feasible. To mitigate data scarcity and enhance accessibility, we investigate the generation of synthetic DTI scalar maps--specifically mean diffusivity (MD)--from structural 3D volumetric T1-weighted brain MRI using a reversible generative adversarial network (RevGAN). Unlike conventional pipelines requiring multiple steps, our approach enables a single-step translation from T1 to diffusion-derived measures. We assess the quality and utility of the synthetic maps in two downstream tasks: sex classification and Alzheimers disease classification. Performance comparisons between models trained on real and synthetic DTI maps demonstrate that RevGAN-generated images retain meaningful microstructural features and offer competitive accuracy, underscoring their potential for data augmentation and analysis in neuroimaging workflows. We also examine how well models trained on these data generalize to a new population dataset from India (NIMHANS cohort).

neuroscience↗

bioRxiv2025

Proximity-based super-resolution imaging enabled by DNA base-stacking interactions

Super-resolution imaging has gained significant traction in recent years due to its unprecedented ability to visualize target biomolecules at nanometer resolution. Here, we demonstrate the capability to detect target pairs that are in close proximity by exploiting base-stacking interactions between two DNA strands, each labelling one target. Our DNA probes hybridize transiently with the each other only when both target molecules are proximal, thus creating a hybridization site for fluorophore-conjugated DNA strand, called imager. In this design, hybridization and co-axial base-stacking act synergistically to enable imager binding, with stacking interactions providing essential stabilization that allows for transient hybridization. This synergy generates the stochastic binding events required for DNA-PAINT imaging, which we call Stack-Proximity-PAINT (Stack-pPAINT). To gain mechanistic insights into hybridization of our probe, we performed atomistic equilibrium and steered molecular dynamics (MD) simulations. The simulations reveal that DNA base-stacking and fluorophore stacking together stabilize the imager. We utilized programmable DNA nanostructures to benchmark the applicability of Stack-pPAINT. As a cellular proof of concept, we visualized microtubular structures using Stack-pPAINT with antibodies targeting both alpha- and beta-tubulin molecules. This probe technology offers promising applications in cell biology research aimed at elucidating spatial interactomes at high resolution within cells.

biophysics↗

bioRxiv2024

Lifespan Normative Modeling of Brain Microstructure

Normative models of brain metrics based on large populations could be extremely valuable for detecting brain abnormalities in patients with a variety of disorders, including degenerative, psychiatric and neurodevelopmental conditions, but no such models exist for the brains white matter (WM) microstructure. Here we present a large-scale normative model of brain WM microstructure - based on 19 international diffusion MRI datasets covering almost the entire lifespan (totaling N=54,583 individuals; age: 4-91 years). We extracted regional diffusion tensor imaging (DTI) metrics using a standardized analysis and quality control protocol and used hierarchical Bayesian regression (HBR) to model the statistical distribution of derived WM metrics as a function of age and sex. We extracted the average lifespan trajectories and corresponding centile curves for each WM region. We illustrate the utility of the method by applying it to detect and visualize profiles of WM microstructural deviations in a variety of contexts: in mild cognitive impairment, Alzheimers disease, and 22q11.2 deletion syndrome - a neurogenetic condition that markedly increases risk for schizophrenia. The resulting large-scale model provides a common reference to identify disease effects on the brains microstructure in individuals or groups, and to compare disorders, and discover factors affecting WM abnormalities. The derived normative models are a valuable resource publicly available to the community, adaptable and extendable to future datasets as the available data expands.

neuroscience↗

bioRxiv2024

Comparison of Explainable AI Models for MRI-based Alzheimer's Disease Classification

Deep learning models based on convolutional neural networks (CNNs) have been used to classify Alzheimers disease or infer dementia severity from 3D T1-weighted brain MRI scans. Here, we examine the value of adding occlusion sensitivity analysis (OSA) and gradient-weighted class activation mapping (Grad-CAM) to these models to make the results more interpretable. Much research in this area focuses on specific datasets such as the Alzheimers Disease Neuroimaging Initiative (ADNI) or National Alzheimers Coordinating Center (NACC), which assess people of North American, predominantly European ancestry, so we examine how well models trained on these data generalize to a new population dataset from India (NIMHANS cohort). We also evaluate the benefit of using a combined dataset to train the CNN models. Our experiments show feature localization consistent with knowledge of AD from other methods. OSA and Grad-CAM resolve features at different scales to help interpret diagnostic inferences made by CNNs.

neuroscience↗

bioRxiv2024

Microstructural Mapping of Neural Pathways in Alzheimer's Disease using Macrostructure-Informed Normative Tractometry

IntroductionDiffusion MRI is sensitive to the microstructural properties of brain tissues, and shows great promise in detecting the effects of degenerative diseases. However, many approaches analyze single measures averaged over regions of interest, without considering the underlying fiber geometry. MethodsHere, we propose a novel Macrostructure-Informed Normative Tractometry (MINT) framework, to investigate how white matter microstructure and macrostructure are jointly altered in mild cognitive impairment (MCI) and dementia. We compare MINT-derived metrics with univariate metrics from diffusion tensor imaging (DTI), to examine how fiber geometry may impact interpretation of microstructure. ResultsIn two multi-site cohorts from North America and India, we find consistent patterns of microstructural and macrostructural anomalies implicated in MCI and dementia; we also rank diffusion metrics sensitivity to dementia. DiscussionWe show that MINT, by jointly modeling tract shape and microstructure, has potential to disentangle and better interpret the effects of degenerative disease on the brains neural pathways.

neuroscience↗

bioRxiv2024

Brain Age Analysis and Dementia Classification using Convolutional Neural Networks trained on Diffusion MRI: Tests in Indian and North American Cohorts

Deep learning models based on convolutional neural networks (CNNs) have been used to classify Alzheimers disease or infer dementia severity from T1-weighted brain MRI scans. Here, we examine the value of adding diffusion-weighted MRI (dMRI) as an input to these models. Much research in this area focuses on specific datasets such as the Alzheimers Disease Neuroimaging Initiative (ADNI), which assesses people of North American, largely European ancestry, so we examine how models trained on ADNI, generalize to a new population dataset from India (the NIMHANS cohort). We first benchmark our models by predicting "brain age" - the task of predicting a persons chronological age from their MRI scan and proceed to AD classification. We also evaluate the benefit of using a 3D CycleGAN approach to harmonize the imaging datasets before training the CNN models. Our experiments show that classification performance improves after harmonization in most cases, as well as better performance for dMRI as input.

neuroscience↗

bioRxiv2024

Deep Normative Tractometry for Identifying Joint White Matter Macro- and Micro-structural Abnormalities in Alzheimer's Disease

This study introduces the Deep Normative Tractometry (DNT) framework, that encodes the joint distribution of both macrostructural and microstructural profiles of the brain white matter tracts through a variational autoencoder (VAE). By training on data from healthy controls, DNT learns the normative distribution of tract data, and can delineate along-tract micro- and macro-structural abnormalities. Leveraging a large sample size via generative pre-training, we assess DNTs generalizability using transfer learning on data from an independent cohort acquired in India. Our findings demonstrate DNTs capacity to detect widespread diffusivity abnormalities along tracts in mild cognitive impairment and Alzheimers disease, aligning closely with results from the Bundle Analytics (BUAN) tractometry pipeline. By incorporating tract geometry information, DNT may be able to distinguish disease-related abnormalities in anisotropy from tract macrostructure, and shows promise in enhancing fine-scale mapping and detection of white matter alterations in neurodegenerative conditions.

neuroscience↗

bioRxiv2023

In silico and in vitro studies suggest epigallocatechin gallate (EGCG), a polyphenol in green tea, can bind and modulate the aggregation and cytotoxicity of the full-length TDP-43 protein implicated in TDP-43 proteinopathies.

Misfolding and aggregation of TDP-43 protein are implicated in several proteinopathies like ALS and FTLD. Extracellular TDP-43 is also proposed to propagate in a prion-like pathogenic manner to the neighbouring cells. Here, using turbidity and sedimentation assay, we show that a polyphenol in green tea, epigallocatechin gallate (EGCG), can inhibit the in vitro aggregation of the full-length TDP-43 protein. Furthermore, Alexa-Fluor-labelled TDP-43 protein failed to show aggregates in the presence of EGCG in fluorescence microscopy. Also, AFM imaging revealed that EGCG co-incubation with TDP-43 allows formation of only small oligomers in contrast to the larger TDP-43 aggregates formed otherwise. A physical binding of EGCG with TDP-43 was observed using triphenyl tetrazolium chloride (TTC) staining and isothermal titration calorimetry (ITC). ITC also revealed a high-affinity binding site for EGCG on TDP-43 with a Kd value of 7.8 {micro}M and a binding free energy of -6.9 kcal/mol. Furthermore, in silico molecular docking and molecular dynamic simulation (MDS) studies using different available structures of the N-terminal, RRM1-2 and C-terminal domains of TDP-43, predicted a preferable and stable binding of EGCG to the structure of the aggregation prone C-terminal domain (CTD) (PDB ID:7KWZ). Also, EGCG complexed with CTD of TDP-43 yielded a negative {Delta}G value of -20.29 kcal/mol using MM-PBSA analysis of the MDS data thereby further suggesting a stable complex formation. Also, in MDS, EGCG interacted with the amino acids Phe-313 and Ala-341 of TDP-43, which were previously projected to be important for the recruitment of monomers for the amyloid formation by CTD, thereby suggesting a possible mechanism of EGCGs inhibition of the TDP-43 aggregation. Notably, while the in vitro-made aggregates of full-length TDP-43 caused mild cytotoxicity to the HEK293 cells, the small oligomers of TDP-43 formed in presence of EGCG did not. In totality, EGCG can in vitro interact with TDP-43 and inhibit its aggregation, possibly via interaction with the amyloidogenic domain, thereby preventing it from assuming cytotoxic conformations. As EGCG is a natural molecule, it could be relevant to the therapeutic quest against the TDP-43 proteinopathies.

biochemistry↗

bioRxiv2023

High-throughput and scalable single cell proteomics identifies over 5000 proteins per cell

The emergence of mass spectrometry (MS)-based single-cell proteomics (SCP) promise to revolutionize the study of cellular biology and biomedicine by providing an unparalleled view of the proteome in individual cells. Despite its groundbreaking potential, SCP is nascent and faces challenges including limited sequence depth, throughput, and reproducibility, which have constrained its broader utility. This study introduces key methodological advances, which considerably improve the sensitivity, coverage and dependability of protein identification from single cells. We developed an almost lossless SCP workflow encompassing sample preparation to MS analysis, doubling the number of identified proteins from roughly 2000 to over 5000 in individual HeLa cells. A comprehensive evaluation of analytical software tools, alongside strict false discovery rate (FDR) controls solidified the reliability of our results. These enhancements also facilitated the direct detection of post-translational modifications (PTMs) in single cells, negating the need for enrichment and thereby simplifying the analytical process. Although throughput in MS remains a challenge, our study demonstrates the feasibility of processing up to 80 label-free SCP samples per day. Moreover, an optimized tissue dissociation buffer enabled effective single cell disaggregation of drug-treated cancer cell spheroids, refining the overall proteomic analysis. Our workflow sets a new benchmark in SCP for sensitivity and throughput, with broad applications ranging from the study of cellular development to disease progression and the identification of cell type-specific markers and therapeutic targets.

systems biology↗

bioRxiv2023

Generation of antigen-specific memory CD4 T cells by heterologous immunization enhances the magnitude of the germinal center response upon influenza infection

Current influenza vaccine strategies have yet to overcome significant obstacles, including rapid antigenic drift of seasonal influenza viruses, in generating efficacious long-term humoral immunity. Due to the necessity of germinal center formation in generating long-lived high affinity antibodies, the germinal center has increasingly become a target for the development of novel or improvement of less-efficacious vaccines. However, there remains a major gap in current influenza research to effectively target T follicular helper cells during vaccination to alter the germinal center reaction. In this study, we used a heterologous infection or immunization priming strategy to seed an antigen-specific memory CD4+ T cell pool prior to influenza infection in mice to evaluate the effect of recalled memory T follicular helper cells in increased help to influenza-specific primary B cells and enhanced generation of neutralizing antibodies. We found that heterologous priming with intranasal infection with acute lymphocytic choriomeningitis virus (LCMV) or intramuscular immunization with adjuvanted recombinant LCMV glycoprotein induced increased antigen-specific effector CD4+ T and B cellular responses following infection with a recombinant influenza strain that expresses LCMV glycoprotein. Heterologously primed mice had increased expansion of secondary Th1 and Tfh cell subsets, including increased CD4+ TRM cells in the lung. However, the early enhancement of the germinal center cellular response following influenza infection did not impact influenza-specific antibody generation or B cell repertoires compared to primary influenza infection. Overall, our study suggests that while heterologous infection/immunization priming of CD4+ T cells is able to enhance the early germinal center reaction, further studies to understand how to target the germinal center and CD4+ T cells specifically to increase long-lived antiviral humoral immunity are needed. Author SummaryT follicular helper (Tfh) cells are specialized CD4+ T cells that provide help to B cells and are required to form germinal centers within secondary lymphoid organs during an immune response. Germinal centers are necessary for generating high affinity virus-specific antibodies necessary to clear influenza infections, though current vaccines fail to generate long-lived antibodies that universally recognize different influenza strains. We used a "heterologous priming" strategy in mice using a non-influenza viral infection or viral protein subunit vaccination to form memory CD4+ Tfh cells (in previously naive mice) that can be rapidly recalled into secondary Tfh cells following influenza infection and ideally enhance the germinal center reaction and formation of high affinity antibodies to influenza better than primary Tfh cells. Our study showed that heterologous priming induced an increase in both CD4+ T and B cells early following influenza infection, suggesting we could successfully target enhancement of the germinal center. Despite the enhancement of the early germinal center cellular response, we did not see an increase in influenza-specific antiviral antibodies. Thus, while Tfh cells are critical for the generation of high affinity antibodies, other strategies to target expansion of Tfh cells during influenza vaccination will need to be developed.

immunology↗

bioRxiv2023

BundleCleaner: Unsupervised Denoising and Subsampling of Diffusion MRI-Derived Tractography Data

We present BundleCleaner, an unsupervised multi-step frame-work that can filter, denoise and subsample bundles derived from diffusion MRI-based whole-brain tractography. Our approach considers both the global bundle structure and local streamline-wise features. We apply BundleCleaner to bundles generated from single-shell diffusion MRI data in an independent clinical sample of older adults from India using probabilistic tractography and the resulting cleaned bundles can better align with the atlas bundles with reduced overreach. In a downstream tractometry analysis, we show that the cleaned bundles, represented with less than 20% of the original set of points, can robustly localize along-tract microstructural differences between 32 healthy controls and 34 participants with Alzheimers disease ranging in age from 55 to 84 years old. Our approach can help reduce memory burden and improving computational efficiency when working with tractography data, and shows promise for large-scale multi-site tractometry.

neuroscience↗

bioRxiv2023

Interpretation of Predictions in Drug-Gut Bacteria Interactions Using Machine Learning

Gut bacteria play a crucial role in hosts metabolism. Both antibiotic and non-antibiotic drugs affect the gut bacteria ecosystem, which negatively affects the hosts health. Also, gut bacteria metabolize drugs, making them ineffective to the target. The structure-activity relationship studies of drugs have the scope to make them more effective, efficient, and specific to the target. Previous machine learning studies use the available data to predict the activity of drugs and gut bacteria on each other, but these models lack interpretability. Herein, we study the drug-gut bacteria interaction using interpretable machine learning models. In this study, we identify the most important physicochemical features of the drug, which decide the drug-gut bacteria interactions with each other. One of the key findings of this work is that the higher-positive charged drug molecules can inhibit the growth of gut bacteria more. In contrast, the higher-negative charged drug molecules have higher possibility to get metabolized by gut bacteria.

biophysics↗