Biology subjects

Chaudhary, K.

Publications and source records attributed to Chaudhary, K..

11 recordsLinked to original sources

bioRxiv2026

A Comparative Benchmark of Biomedical Language Models for Concept Normalization from Real-World Text

Patient-reported and clinically documented narratives often contain informal, fragmented and linguistically heterogeneous expressions, complicating medical concept normalization (MCN) and frequently necessitating pre-processing before terminology mapping. Despite rapid advances in biomedical language modeling, the comparative utility of representation models for MCN and their integration with instruction-tuned LLMs in automated normalization workflows remain underexplored. In this study, we systematically benchmarked 15 general, biomedical and clinical transformer-based representation models together with 12 instruction-tuned LLMs across 12,713 instances from five established datasets: TAC2017_ADR, TwADR-L, TwiMed, CADEC and SMM4H2017. Representation models were evaluated using embedding-based semantic retrieval, whereas instruction-tuned LLMs were assessed as upstream text-correction modules. SapBERT achieved the highest Top-5 accuracy among representation models, reaching 63.8% for SNOMED CT and 58.0% for MedDRA without correction. Qwen 2 Instruct was selected as the preferred corrector on the basis of its favorable balance between Top-1 normalization performance and computational efficiency relative to substantially larger models, including Llama 3.1 Instruct (70B). Incorporation of Qwen 2 Instruct increased Top-5 accuracy to 69.4% for SNOMED CT and 63.2% for MedDRA. The resulting framework accepts heterogeneous short medical expressions without manual input pre-processing and automatically performs text refinement, semantic retrieval and terminology mapping to standardized concepts and vocabulary codes. These findings establish a benchmark-guided, scalable framework for automated medical terminology standardization.

bioinformatics↗

bioRxiv2025

PEPlife2: A Updated Repository of the Half-life of Peptides

This manuscript presents an updated version of PEPlife, a manually curated database that provides comprehensive information on peptide half-life. The updated version, PEPlife2, includes 4,412 entries, compared to 2,229 entries in the previous version. Each entry contains detailed information that include experimental methods used to determine half-life, chemical modifications, biological activity, routes of administration, and other relevant data. The database encompasses 402 proteins and 1,781 peptides, derived from 105 unique assays. In addition to natural peptide sequences, PEPlife2 includes cyclic peptides and chemically modified peptides, such as those with N- and C-terminal modifications. To provide structural insights, peptide and protein structures are sourced from the Protein Data Bank (PDB) or predicted using PEPstrMOD. To support the scientific community, we have developed a user-friendly interface and integrated advanced analytical tools, including BLAST, Smith-Waterman, GGSEARCH, CLUSTALW, and MUSTANG. The updated database is accessible at: https://webs.iiitd.edu.in/raghava/peplife2/. Authors BiographyO_LIUrooj Alam is currently pursuing a Masters degree in Computational Biology at the Department of Computational Biology, Indraprastha Institute of Information Technology, New Delhi, India. C_LIO_LIKunal Chaudhary is working as Ph.D. in Computational biology from Department of Computational Biology, Indraprastha Institute of Information Technology, New Delhi, India C_LIO_LINishant Kumar is currently working as Ph.D. in Computational biology from Department of Computational Biology, Indraprastha Institute of Information Technology, New Delhi, India C_LIO_LIRitu Tomer is currently working as Ph.D. in Computational biology from Department of Computational Biology, Indraprastha Institute of Information Technology, New Delhi, India C_LIO_LISumeet Patiyal is currently working as a postdoctoral visiting fellow Cancer Data Science Laboratory, National Cancer Institute, National Institutes of Health, Bethesda, Maryland, USA. C_LIO_LIGajendra P. S. Raghava is currently working as Professor and Head of Department of Computational Biology, Indraprastha Institute of Information Technology, New Delhi, India. C_LI

bioinformatics↗

bioRxiv2021

Identification of Discriminative Gene-level and Protein-level Features Associated with Gain-of-Function and Loss-of-Function Mutations

Identifying whether a given genetic mutation results in a gene product with increased (gain-of-function; GOF) or diminished (loss-of-function; LOF) activity is an important step toward understanding disease mechanisms as they may result in markedly different clinical phenotypes. Here, we generated the first extensive database of all currently known germline GOF and LOF pathogenic mutations by employing natural language processing (NLP) on the available abstracts in the Human Gene Mutation Database. We then investigated various gene- and protein-level features of GOF and LOF mutations by applying machine learning and statistical analyses to identify discriminative features. We found that GOF mutations were enriched in essential genes, autosomal dominant inheritance, protein binding and interaction domains, whereas LOF mutations were enriched in singleton genes, protein-truncating variants, and protein core regions. We developed a user-friendly web-based interface that enables the extraction of selected subsets from the GOF/LOF database by a comprehensive set of annotated features, and downloading up-to-date versions (https://itanlab.shinyapps.io/goflof/). These results could ultimately improve our understanding of how mutations affect gene/protein function thereby guiding future treatment options.

genetics↗

bioRxiv2020

Isolation and characterization of lumpy skin disease virus from cattle in India

Lumpy skin disease (LSD) has devastating economic impact. During the last decade, LSD had spread to climatically new and previously disease-free countries, which also includes its recent emergence in the Indian subcontinent (2019). This study deals with the LSD outbreak(s) from cattle in Ranchi (India). Virus was isolated from the scabs (skin lesions) in the primary goat kidney cells. Phylogenetic analysis based on nucleotide sequencing of LSD virus (LSDV) ORF011, ORF012 and ORF036 suggested that the isolated virus (LSDV/Bos taurus-tc/India/2019/Ranchi) is closely related to Kenyan LSDV strains. Further, we adapted the isolated virus in Vero cells, an alternative cell line that was found suitable for LSDV propagation. In addition, kinetics of viral DNA synthesis and one-step growth curve analysis suggested that LSDV initiates synthesizing its genome at [~]24 hours post-infection (hpi) with a peak level at [~]96 hpi whereas evidence of progeny virus particles was observed at 36-48 hours (h) with a peak titre at [~]120 h. This study reports the first successful isolation of LSDV in India, besides describing alternative cells for virus isolation (primary goat kidney cells) and in vitro propagation (Vero cells) as well as providing some insights on LSDV life cycle. Author SummaryThis study describes the first successful isolation of LSDV in India. Nucleotide sequencing of LSDV ORF011, ORF012 and ORF036 suggested that the isolated virus is closely related to Kenyan LSDV strains. This is in accordance with the previous (only) study, thereby suggesting that a single LSDV strain is circulating in India. For the first time, we employed primary goat kidney (PGK) cells for the isolation of LSDV, where PGK cells were found more sensitive for LSDV isolation as compared to the primary lamb testicle (PLT) and primary goat testicle (PGT) cells. We also adapted the isolated virus in Vero cells and demonstrated that Vero cells are also suitable for in vitro propagation of LSDV. Besides, we also provide some insights on the LSDV life cycle.

microbiology↗

bioRxiv2020

Computational Support, Not Primacy, Distinguishes Compensatory Memory Reorganization in Epilepsy

Temporal lobe epilepsy (TLE) is associated with impairment in episodic memory. A substantial sub-group, however, is able to maintain adequate memory despite temporal lobe pathology. What has been missing from prior work in the cognitive reorganization is a direct comparison of TLE patients with intact/compensated status from those who are memory impaired and uncompensated. Little is known about the particular regional activations, functional connectivities (FCs), and/or network reconfigurations that implement changes in the primary computations or the support functions that drive adaptive plasticity and compensated memory. We utilized task fMRI on 54 unilateral TLE patients and 24 matched healthy controls (HC) during performance of a paired-associate memory (PAM) task to address three questions: 1) what regions implement PAM in TLE, and do such regions vary as a function of good versus poor performance, 2) are there unique FCs present during memory encoding that accounts for intact status through the preservation of primary memory computations or the supportive computations that allow for compensated memory responses, and 3) what features during memory encoding are most distinctive: is it the magnitude and location of regional activations, or the presence of enhanced functional connections to key structures such as the hippocampus? Results revealed a unique profile of non-ictal, non-dominant hemisphere regions (e.g., right posterior temporal regions) were most important to intact/compensated status in LTLE, involving both increased regional activity and increased modulatory communication with the hippocampi, all feature that was missing in impaired/uncompensated LTLE. The profile involved areas that are neither contralateral homologues to left hemisphere memory areas, nor regions traditionally considered computationally primary for episodic memory. None of these areas of increased activation or functional connectivity were associated with advantaged memory in HCs. Our emphasis on different performance levels yielded insight into two forms of cognitive reorganization. Computational primacy, where LTLE showed little change relative to HCs, and computational support where Intact/Compensated LTLE patients showed adaptive abnormalities. The analyses isolated the unique regional activations and mediating FCs that implement truly compensatory reorganization in LTLE. The results provided a new perspective on memory deficits by making clear that memory deficits arise not just from knockout of a functional hub, but from the failure to instantiate a complex set of reorganization responses. Such responses provided computational support to ensure successful memory. The findings demonstrated that by keeping track of performance levels, we can increase our understanding of adaptive brain responses and neuroplasticity in epilepsy.

neuroscience↗

bioRxiv2018

Unsupervised Machine learning to subtype Sepsis-Associated Acute Kidney Injury

ObjectiveAcute kidney injury (AKI) is highly prevalent in critically ill patients with sepsis. Sepsis-associated AKI is a heterogeneous clinical entity, and, like many complex syndromes, is composed of distinct subtypes. We aimed to agnostically identify AKI subphenotypes using machine learning techniques and routinely collected data in electronic health records (EHRs).\n\nDesignCohort study utilizing the MIMIC-III Database.\n\nSettingICUs from tertiary care hospital in the U.S.\n\nPatientsPatients older than 18 years with sepsis and who developed AKI within 48 hours of ICU admission.\n\nInterventionsUnsupervised machine learning utilizing all available vital signs and laboratory measurements.\n\nMeasurements and Main ResultsWe identified 1,865 patients with sepsis-associated AKI. Ten vital signs and 691 unique laboratory results were identified. After data processing and feature selection, 59 features, of which 28 were measures of intra-patient variability, remained for inclusion into an unsupervised machine-learning algorithm. We utilized k-means clustering with k ranging from 2 - 10; k=2 had the highest silhouette score (0.62). Cluster 1 had 1,358 patients while Cluster 2 had 507 patients. There were no significant differences between clusters on age, race or gender. We found significant differences in comorbidities and small but significant differences in several laboratory variables (hematocrit, bicarbonate, albumin) and vital signs (systolic blood pressure and heart rate). In-hospital mortality was higher in cluster 2 patients, 25% vs. 20%, p=0.008. Features with the largest differences between clusters included variability in basophil and eosinophil counts, alanine aminotransferase levels and creatine kinase values.\n\nConclusionsUtilizing routinely collected laboratory variables and vital signs in the EHR, we were able to identify two distinct subphenotypes of sepsis-associated AKI with different outcomes. Variability in laboratory variables, as opposed to their actual value, was more important for determination of subphenotypes. Our findings show the potential utility of unsupervised machine learning to better subtype AKI.

bioinformatics↗

bioRxiv2018

A web bench for analysis and prediction of oncological status from proteomics data of urine samples

Urine-based cancer biomarkers offer numerous advantages over the other biomarkers and play a crucial role in cancer management. In this study, an attempt has been made to develop proteomics-based prediction models to discriminate patients of oncological disorders related to urinary tract and healthy controls from their urine samples. The dataset used in this study was obtained from human urinary peptide database that contains urine proteomics data of 1525 oncological and 1503 healthy controls with the spectral intensity of 5605 peptides. First, we identified peptide spectra using various feature selection techniques, which display different intensity and occurrence in oncological samples and healthy controls. Based on selected 173 peptide-based biomarkers, we developed models for predicting oncological samples and achieved maximum accuracy of 91.94% with 0.84 MCC. Prediction models were also developed based on spectral intensities with known peptide sequences. We also quantitated the amount of protein in a sample based on intensities of its fragments/peptides and developed prediction models based on protein expression. It was observed that certain proteins and their peptides such as fragments of collagen protein are more abundant in oncological samples. Based on this study, we also developed a web bench, CancerUBM, for mining proteomics data, which is freely available at http://webs.iiitd.edu.in/raghava/cancerubm/.

bioinformatics↗

bioRxiv2017

Computer-aided prediction of antigen presenting cell modulators for designing peptide-based vaccine adjuvants

BackgroundEvidences in literature strongly advocate the potential of immunomodulatory peptides for use as vaccine adjuvants. All the mechanisms of vaccine adjuvants ensuing immunostimulatory effects directly or indirectly stimulate Antigen Presenting Cells (APCs). While numerous methods have been developed in the past for predicting B-cell and T-cell epitopes; no method is available for predicting the peptides that can modulate the APCs.\n\nMethodsWe named the peptides that can activate APCs as A-cell epitopes and developed methods for their prediction in this study. A dataset of experimentally validated A-cell epitopes was collected and compiled from various resources. To predict A-cell epitopes, we developed Support Vector Machine-based machine learning models using different sequence-based features.\n\nResultsA hybrid model developed on a combination of sequence-based features (dipeptide composition and motif occurrence), achieved the highest accuracy of 96.91% with Matthews Correlation Coefficient (MCC) value of 0.94 on the training dataset. We also evaluated the hybrid models on an independent dataset and achieved a comparable accuracy of 94.93% with MCC 0.90.\n\nConclusionThe models developed in this study were implemented in a web-based platform VaxinPAD to predict and design immunomodulatory peptides or A-cell epitopes. This web server available at http://webs.iiitd.edu.in/raghava/vaxinpad/ and http://crdd.osdd.net/raghava/vaxinpad/ will facilitate researchers in designing peptide-based vaccine adjuvants.

bioinformatics↗

bioRxiv2017

Deep learning accurately predicts estrogen receptor status in breast cancer metabolomics data

Metabolomics holds the promise as a new technology to diagnose highly heterogeneous diseases. Conventionally, metabolomics data analysis for diagnosis is done using various statistical and machine learning based classification methods. However, it remains unknown if deep neural network, a class of increasingly popular machine learning methods, is suitable to classify metabolomics data. Here we use a cohort of 271 breast cancer tissues, 204 positive estrogen receptor (ER+) and 67 negative estrogen receptor (ER-), to test the accuracies of autoencoder, a deep learning (DL) framework, as well as six widely used machine learning models, namely Random Forest (RF), Support Vector Machines (SVM), Recursive Partitioning and Regression Trees (RPART), Linear Discriminant Analysis (LDA), Prediction Analysis for Microarrays (PAM), and Generalized Boosted Models (GBM). DL framework has the highest area under the curve (AUC) of 0.93 in classifying ER+/ER-patients, compared to the other six machine learning algorithms. Furthermore, the biological interpretation of the first hidden layer reveals eight commonly enriched significant metabolomics pathways (adjusted P-value<0.05) that cannot be discovered by other machine learning methods. Among them, protein digestion & absorption and ATP-binding cassette (ABC) transporters pathways are also confirmed in integrated analysis between metabolomics and gene expression data in these samples. In summary, deep learning method shows advantages for metabolomics based breast cancer ER status classification, with both the highest prediction accurcy (AUC=0.93) and better revelation of disease biology. We encourage the adoption of autoencoder based deep learning method in the metabolomics research community for classification.

bioinformatics↗

bioRxiv2017

Phenotypic associations of consensus driver genes in hepatocellular carcinoma

Although driver genes in hepatocellular carcinoma (HCC) have been investigated in various previous genetic studies, prevalence of key driver genes among heterogeneous populations is unknown. Moreover, the phenotypic associations of these driver genes are poorly understood. This report aims to reveal the phenotypic impacts of a group of consensus driver genes in HCC. We used MutSigCV and OncodriveFM modules implemented in the IntOGen pipeline to identify consensus driver genes across six HCC cohorts comprising 1,494 samples in total. To access their global impacts, we used TCGA mutations and copy number variations to predict the transcriptomics data, under generalized linear models. We further investigated the associations of the consensus driver genes to patient survival, age, gender, race and risk factors. We identify 10 consensus driver genes across six HCC cohorts in total. Integrative analysis of driver mutations, copy number variations and transcriptomic data reveals that these consensus driver mutations and their copy number variations are associated with majority (62.5%) of the mRNA transcriptome, but only a small fraction (8.9%) of miRNAs. Genes associated with TP53, CTNNB1, and ARID1A mutations contribute to the tripod of most densely connected pathway clusters. These driver genes are significantly associated with patients overall survival. Some driver genes are significantly linked to HCC gender (CTNNB1, ALB, TP53 and AXIN1), race (TP53 and CDKN2A), and age (RB1) disparities. This study prioritizes a group of consensus drivers in HCC, which collectively show vast impacts on the phenotypes. These driver genes may warrant as valuable therapeutic targets of HCC.

bioinformatics↗

bioRxiv2017

Deep Learning based multi-omics integration robustly predicts survival in liver cancer

Identifying robust survival subgroups of hepatocellular carcinoma (HCC) will significantly improve patient care. Currently, endeavor of integrating multi-omics data to explicitly predict HCC survival from multiple patient cohorts is lacking. To fill in this gap, we present a deep learning (DL) based model on HCC that robustly differentiates survival subpopulations of patients in six cohorts. We build the DL based, survival-sensitive model on 360 HCC patients data using RNA-seq, miRNA-seq and methylation data from TCGA, which predicts prognosis as good as an alternative model where genomics and clinical data are both considered. This DL based model provides two optimal subgroups of patients with significant survival differences (P=7.13e-6) and good model fitness (C-index=0.68). More aggressive subtype is associated with frequent TP53 inactivation mutations, higher expression of stemness markers (KRT19, EPCAM) and tumor marker BIRC5, and activated Wnt and Akt signaling pathways. We validated this multi-omics model on five external datasets of various omics types: LIRI-JP cohort (n=230, C-index=0.75), NCI cohort (n=221, C-index=0.67), Chinese cohort (n=166, C-index=0.69), E-TABM-36 cohort (n=40, C-index=0.77), and Hawaiian cohort (n=27, C-index=0.82). This is the first study to employ deep learning to identify multi-omics features linked to the differential survival of HCC patients. Given its robustness over multiple cohorts, we expect this workflow to be useful at predicting HCC prognosis prediction.

bioinformatics↗