Debarka Sengupta

The paper introduces kGain, a metric quantifying k-mer frequency shifts due to mutations, and shows it predicts adaptive potential across species and experimental systems.

scKNIFE integrates gene expression with functional annotations to identify tumor programs and treatment-related functional variation in breast cancer.

NucXplore converts H&E-stained nuclei into 129 explicit features, outperforming classical and deep learning methods in histopathology cohorts and revealing aging and exposure-related cellular changes.

The paper introduces BioChirp, a system for reliable retrieval of structured biomedical data, outperforming existing agentic NL2SQL systems in accuracy and consistency.

MutAIverse detects known and novel DNA adducts and their precursor genotoxins using an AI-driven platform with mechanistic modeling and spectral mapping.

The paper identifies endogenous sterols as intracellular allosteric modulators of the GPCR-Gα interface using a deep learning toolkit and experimental validation.

The paper introduces CDI, a framework that unifies multiple molecular representations into a single embedding, achieving better predictive performance than individual methods.

This paper introduces FORGE, a framework that jointly models drug response and gene essentiality to identify treatment groups and predict efficacy using basal gene expression.

SynGlue is an AI framework that designs multi-target therapeutics by combining data-driven and structure-guided methods, producing potent degraders for BRD4 and GSPT1.

RCoxNet improves cancer survival prediction by integrating mutation and clinical data with a deep learning Cox model and random walk algorithm, achieving better performance than existing models.

The paper introduces a Python package for DNA sequence encoding using Frequency Chaos Game Representation (FCGR), showing improved accuracy and efficiency over an R-based library for DNA sequences.

InGene identifies influential genes in scRNA-seq data by assigning importance scores based on their contribution to nonlinear dimensionality reduction maps.

scEpiSearch enables accurate matching of single-cell open-chromatin profiles to large reference datasets, improving co-embedding and classification of epigenomic states across species and platforms.

The paper introduces unCTC, an R package for unbiased identification and characterization of CTCs from scRNA-Seq data, validated on breast CTCs.

SCellBOW identifies and risk-stratifies tumor cell subpopulations using scRNA-seq data by simulating their impact on disease prognosis.

Stardust is a visualization method that embeds cells and marker genes together on a 2D map, performing competitively with existing tools on scRNA-seq and spatial transcriptomics data.

The paper develops Metabokiller, an AI model that predicts carcinogenic human metabolites by analyzing their biochemical properties and validates predictions through experimental assays.

The paper constructs a visual atlas of gene-pathology relationships by mining PubMed abstracts and using NLP to identify tissue-specific gene roles in disease.

ROSeq provides a robust differential expression analysis method for scRNA-seq data by modeling expression ranks, showing improved noise tolerance and scalability.

UniPath uses pathway and gene-set enrichment scores to analyze single-cell epigenome and transcriptome data, revealing biologically meaningful co-clustering and aiding in cancer cell regulation inference.

The paper analyzes single-cell expression profiles of circulating tumor cells (CTCs) across cancers, identifies EMT-related patterns, and develops a classifier to recognize CTCs using PBMC transcriptomes.

This study uses deep learning to analyze exome sequencing data and detect cancer-specific SNVs, addressing the challenge of mutation exclusivity in cancer.

deepMc uses deep matrix factorization to impute missing values in single cell RNA-seq data, improving clustering, differential expression analysis, and cell type separability.

dropClust2 is an R package that improves the efficiency and resource usage of single cell RNA-seq clustering with a new batch effect removal algorithm.

McImpute improves single cell RNA-seq data by imputing dropouts, enhancing clustering, differential expression, and visualization.

dropClust efficiently clusters ultra-large scRNA-seq data using Locality Sensitive Hashing, outperforming existing methods in speed and accuracy.

FORKS improves pseudo-temporal ordering of single-cell and bulk data by combining k-means and Steiner trees, showing higher accuracy and robustness than existing methods.