Biology subjects

Mitra, P.

Publications and source records attributed to Mitra, P..

At least 19 recordsLinked to original sources

bioRxiv2026

Explainable HGT-based framework for predicting human dark kinase protein-pathway associations by leveraging BERT-based embeddings and WGAN-GP

Discovery of pathway associations and druggability can leverage underutililized dark kinase genes for treating complex diseases (proven for cancer and neurodegeneration), boosted with computational methods. Herein, we employ BERT-based embeddings of proteins and pathways (refined via two-stage transformer and heterogeneous graph transformer) and protein-protein and protein-pathway associations-both positive (curated from databases) and negative (generated using Wasserstein Generative Adversarial Networks with gradient penalty) to train XGBoost and lightGBM classifiers for predicting pathways associated to human dark kinase proteins, with important features unveiled through SHAP analysis. All pathways are clustered and proteins related to same pathway clusters are grouped together (via predicted and positive protein-pathway associations). Selected PCOS-related human dark kinase proteins (with high predicted and existent associations to PCOS pathways) are docked with known PCOS drugs for druggability analysis. Our model attains accuracy, F1-score, specificity, MCC, AUROC and AUPRC of 0.9816, 0.9816, 0.9852, 0.9632, 0.9978 and 0.9982 respectively, supersedes existing work, correctly classifies 97.48% of test data, predicts 62225 pathway associations to above proteins, infers functional similarity of 96 such proteins to human protein(s) and traces nine important positive features. Our model can be used to determine varied functionalities and disease relevance of proteins via predicted pathway associations.

bioinformatics↗

bioRxiv2026

Learning the spatial cell-cell communication network to decode multi-channel signaling and predict network-hub vulnerabilities with MOSANIC

Intercellular signaling governs the central decisions of tissue biology, from proliferation and immune recruitment to metabolic adaptation and cell death, and its dysregulation is a hallmark of disease. What matters most are properties of the signaling network as a whole rather than of individual interactions: the hubs that hold the network together and mark rational points of intervention, the relays through which a signal propagates across intermediate cells to reach partners it does not directly contact, the response of tissue-wide communication to the loss of a single node, and the metabolite-mediated axis that operates alongside secreted-protein signalling. Scoring known ligand-receptor pairs yields a ranked interaction list that captures none of these and excludes metabolite signalling. We present MOSANIC (Multi-mOdal Self-Attention Network for Intercellular Communication), which learns a tissues communication network directly from spatial transcriptomics. MOSANIC represents each tissue as a single heterogeneous graph of cells, genes and metabolites, initialises every node with a frozen foundation-model representation (scVI, ESM-2 and ChemBERTa), and propagates these representations through a self-attention network over biologically typed edges. Supervision is restricted to spatial gene-expression prediction and excludes ligand-receptor annotation, rendering the inferred communication statistically independent of the reference databases used for evaluation. Across five spatial datasets spanning three platforms and two species, MOSANIC attains the highest accuracy on all eight independent ligand-receptor benchmarks (mean AUROC 0.756) against nine established methods, resolves a metabolite-receptor channel statistically orthogonal to the peptide channel, and reconstructs multi-step signal relays that concentrate within a compact rich-core of load-bearing hub genes and cells whose removal fragments the network far beyond a degree-preserving null. In-silico knockout of these hubs recovers experimentally reported phenotypes, and, given no prior oncological input, MOSANIC nominates SCARF1 as a previously unrecognised communication hub in breast cancer whose elevated expression predicts significantly worse survival in an independent cohort after adjustment for tumour stage and age (hazard ratio 1.17, P = 0.043). MOSANIC is released as an open-source Python package (mosanic-ccc).

bioinformatics↗

bioRxiv2026

EpiESM-GA: Resource-Efficient Protein Foundation Model Features for Equitable B-Cell Epitope Prediction

Prediction of B-cell epitopes can assist in reducing costly wet-lab screening in vaccine design, diagnostics, and antibody discovery. However, current predictors often suffer from noisy labels, weak generalization, and structure-dependent workflows. Here we present EO_SCPLOWPIC_SCPLOWESM-GA, an efficient sequenceonly pipeline for linear B-cell epitope prediction. Positive and negative peptide examples are collected from IEDB, which provides experimentally tested epitopes and distinguishes positive and negative epitope records based on assay evidence(Vita et al., 2019). Each peptide is encoded with a frozen ESM-2 protein language model: a bidirectional transformer producing amino acid embeddings for downstream structure and function tasks (Lin et al., 2023). Mean-pooled embeddings are further compressed into a compact 420-feature representation with a genetic algorithm and classified with lightweight Random Forest, XGBoost, or MLP heads. This avoids foundation-model fine-tuning, reduces the number of trainable parameters, improves interpretability, and enables low-resource deployment. On an IEDB-derived benchmark, EO_SCPLOWPIC_SCPLOWESM-GA attains 0.880{+/-} 0.004 AUC-ROC, 0.852{+/-} 0.005 PR-AUC, 82.0 {+/-} 0.6% accuracy, 0.79 {+/-} 0.01 F1, and 0.74{+/-} 0.01 MCC, outperforming dense ESM-2 features and baselines LBCE-XGB, EpitopeVec, and BepiPred-2.0 (mean{+/-} std over five independent random seeds). The framework shows how frozen protein foundation models can enable pandemic preparedness, peptide vaccine prioritization, diagnostic antigen screening, and equitable computational immunology.

bioinformatics↗

bioRxiv2026

HALPred-B: Host-Aware Linear B-Cell Epitope Prediction: Challenges, Limitations, and Variability Across Species

Predicting linear B-cell epitopes is a basic immunoinformatics task that has a direct impact on vaccine design and antibody engineering. Recent advances in machine learning have improved predictive performance, but most existing approaches are trained on aggregated datasets and assume that antigenic patterns are conserved across host organisms. This assumption ignores the immunological variability depending on the host and prevents generalizing the model across species. This is the first systematic host-wise evaluation where we present a systematic machine learning-based analysis of host-aware linear B-cell epitope prediction using curated datasets from the Immune Epitope Database (IEDB). We build separate datasets for human, mouse, and non-human primate hosts and assess several classification models, including Random Forest, Support Vector Machine (SVM), Gradient Boosting, XGBoost, and K-Nearest Neighbors (KNN). The models exploit feature representations derived from sequences, such as AAIndex descriptors, biochemical properties from ExPASy, and dipeptide composition. Our results show that predictive performance differs substantially across hosts. Models achieve up to 86.07% accuracy and 0.93 ROC-AUC on human datasets but lower performance on mouse and non-human primate datasets. This gap underlies dataset bias and sequence distribution differences, as well as the inability of existing features to capture host-specific immunological context. These results indicate that the prediction of linear B-cell epitopes is intrinsically host-specific, and a single global model does not generalize well across species. We propose to incorporate host-aware modeling strategies and organism-specific features for enhanced predictive reliability and biological relevance.

bioinformatics↗

bioRxiv2026

Trait anxiety drives premature disengagement despite intact opportunity-cost sensitivity

Anxiety has been linked to difficulty sustaining engagement with ongoing tasks, even when continued engagement would yield greater rewards, yet the underlying mechanisms remain unclear. Here we examined how trait anxiety shapes sequential foraging decisions using a patch-leaving task grounded in the Marginal Value Theorem (MVT), a normative framework for explore-exploit decisions previously used to reveal altered computations in conditions such as problem gambling and attention-deficit hyperactivity disorder. Across two independent cohorts, participants adjusted patch residence times according to environmental opportunity costs, indicating preserved sensitivity to task structure irrespective of anxiety levels. Despite this, individuals with higher trait anxiety consistently left patches earlier and accrued fewer rewards. Drift diffusion modeling revealed that these deviations arose from reduced reward-driven evidence accumulation, rather than impaired environmental sensitivity or altered learning: trait anxiety reliably reduced the drift rate governing continued exploitation, providing a computational account of premature disengagement. This effect remained robust after accounting for pupil-linked arousal and baseline stress biomarkers, including salivary -amylase and cortisol, which independently constrained decision dynamics. Together, these findings identify reduced reward-evidence accumulation as a core mechanism through which anxiety promotes premature disengagement from rewarding environments.

neuroscience↗

bioRxiv2026

An Artificial Intelligence-based framework for protein interaction design with accelerated KAN-based Positive-Unlabeled learning

Protein design seeks optimal amino acid sequences for target structures, but designing stable protein complexes remains challenging. We introduce a protein interaction design pipeline combining Monte-Carlo simulation with Metropolis-criteria (MCM) and Deep-Learning. It uses Protein-Protein-Interaction(PPI) scores from Deep-Learning-based MaTPIP model to form a PPI-score-based-MCM (PMCM). The work-flow integrates PMCM-driven sequence generation, HDBSCAN-clustering-based selection, and validation via AlphaFold2 and Molecular-Dynamics (MD) simulation. Incorporating learned PPI scores enhances efficiency and feature fusion. A Positive-Unlabeled (PU) learning classifier accelerates sequence validation, while the Kolmogorov-Arnold Network (KAN) improves PU learning over Multi-Layer-Perceptron (MLP). AlphaFold2 predictions yield median Root-Mean-Square-Deviation (RMSD) 1.17[A] , predicted-Template-Modelling (pTM) 0.72, and interface pTM 0.88; 73% of complexes remain within 2[A] RMSD after 100 ns MD, confirming stability. Interface mutations reveal altered interactions. The KAN-based PU model improves F1-score, precision, and AUC by 5%, 11%, and 2% over MLP. Overall, our method outperforms traditional and simulation-based methods while remaining competitive with modern Deep-Learning design frameworks.

bioinformatics↗

bioRxiv2026

Identification of a Novel miRNA Expression Signature for Lung Adenocarcinoma Using Systematic Machine Learning Optimization

Lung adenocarcinoma (LUAD), the most common lung cancer subtype, urgently requires reliable microRNA (miRNA) biomarkers for early detection and therapy. This study introduces a machine learning framework integrating feature stability analysis, precision-recall curves, and resampling strategies (e.g., SMOTE) to robustly identify miRNA signatures from imbalanced TCGA-LUAD data (564 samples: 519 tumor, 45 normal). We selected 8 stable features (hsa-mir-143, hsamir-210, hsa-mir-21, hsa-mir-183, hsa-mir-96, hsa-mir-182, hsa-mir-130b, hsa-mir-141) with 100% cross-fold stability via 10-fold cross-validation. A Random Forest classifier yielded excellent training performance (AUC: 1.0000; accuracy: 98%) and good generalization on an independent test set (AUC: 0.8438; accuracy: 75%). Consistent feature importance across folds supports biological relevance over overfitting. The framework mitigates class imbalance, high dimensionality, and distribution shifts--key hurdles in biomarker discovery. These reproducible miRNAs hold promise as non-invasive diagnostic tools, though external validation underscores generalization challenges across cohorts.

cancer biology↗

bioRxiv2025

A Transformer based method for the Cap Analysis of Gene Expression and Gene Expression Tag associated 5' cap site prediction in RNA

5 RNA capping is one of the major post-transcriptional modifications for the mobility and stability of RNA molecules. Measuring 5 caps of RNAs can help quantify expression levels of mRNAs and lncRNAs. One of the most successful RNAseq methods that have used capping as a tool to quantify expression of transcription is Cap Analysis of Gene Expression(CAGE). Computational prediction of capping can therefore be used as a precursor to the prediction of transcriptional expression. Unfortunately, there is hardly any computational technique that has focused purely on predicting 5 capping. We have developed a transformer-based method for computational prediction of capping from DNA sequences. Our Llama and ReLoRA-based pre-training model, and Llama and LoRA-based fine-tuning model predict 5 cap sites. We have used Leave-one-chromosome-out-cross-validation for our model. The average accuracy, and F1-score after fine-tuning the human genome hg19(mouse genome mm9) for sequence classification is 79.12%(78.09%), and 78.11%(76.17%), respectively. We noted attention peak-based motifs having an aggregate Wilcoxon rank-sum p-value of 1.075e-10 between the attention peak region and the entire context window for the predicted positive motifs; an aggregate p-value of 7.17e-18 for the predicted negative motifs; and an aggregate p-value of 6.70e-08 between the attention peaks of the predicted positive and the predicted negative motifs. Our Llama-based approach aims to create a sequence-based framework to identify 5 capping sites corresponding to CAGE peaks. Our analysis reveals statistically significant motifs from the regions of peak attention scores, which demonstrates biological relevance for some through their resident sites matching with known TF motifs.

bioinformatics↗

bioRxiv2025

DeepPROTECTNeo: A Deep learning-based Personalized and RV-guided Optimization tool for TCR Epitope interaction using Context-aware Transformers

Background: The development of personalized cancer vaccines relies on accurately identifying neoepitopes capable of eliciting strong immune responses. T cell receptor (TCR)-epitope interactions are fundamental to cancer immunotherapy. Traditional computational approaches focus primarily on epitope-major histocompatibility complex (MHC) binding, often overlooking the critical contribution of TCR binding. Furthermore, the clinical applicability of existing methods is constrained by fragmented pipelines that require separate workflows for variant calling, HLA typing, and independent peptide-MHC (pMHC) or peptide-TCR (pTCR) evaluation stages. Results: We present DeepPROTECTNeo, a unified deep learning framework that integrates genomic variant detection, HLA typing, high-affinity pMHC binding prediction, variant-driven TCR repertoire mining, followed by a hybrid transformer-Convolutional Neural Network dual-branch feature extractor with an explicit cross-attention-based deep learning model for TCR-epitope binding prediction. Our reverse vaccinology-inspired biologically informed architecture integrates Bidirectional Long short-term memory (Bi-LSTM) sequence features, convolutional-attention physicochemical/evolutionary descriptors via gated fusion, and TCR numbered contextual embeddings to enable residue-level interpretable modelling. Under a strict TCR-split strategy, it achieved a mean AUROC of 0.7856 and AUPRC of 0.7932 outperforming six state-of-the-art predictors by 4-5% with tight inter-fold stability. The architecture maintains high robustness against structural hard negatives and imbalanced datasets, successfully recovering 18 of 34 validated high-affinity neoepitopes from a patient-specific cancer cohort. Conclusions: Experiments results demonstrate that DeepPROTECTNeo is a powerful, reliable end-to-end neoantigen prioritization framework that effectively models complex TCR-epitope interfaces directly from clinical sequencing data, providing a robust interpretable foundation to accelerate personalized cancer immunotherapy.

bioinformatics↗

bioRxiv2024

E(Q)AGNN-PPIS: Attention Enhanced Equivariant Graph Neural Network for Protein-Protein Interaction Site Prediction

Identifying protein binding sites, the specific regions on a proteins surface where interactions with other molecules occur, is crucial for understanding disease mechanisms and facilitating drug discovery. Although numerous computational techniques have been developed to identify protein binding sites, serving as a valuable screening tool that reduces the time and cost associated with conventional experimental approaches, achieving significant improvements in prediction accuracy remains a formidable challenge. Recent advancements in protein structure prediction, notably through tools like AlphaFold, have made vast numbers of 3-D protein structures available, presenting an opportunity to enhance binding site prediction methods. The availability of detailed 3-D structures has led to the development of Equivariant Graph Neural Networks (GNNs), which can analyze complex spatial relationships in protein structures while maintaining invariance to rotations and translations. However, current equivariant GNN methods still face limitations in fully exploiting the geometric features of protein structures. To address this, we introduce E(Q)AGNN-PPIS 1, an Equivariant Attention-Enhanced Graph Neural Network designed for predicting protein binding sites by leveraging 3-D protein structure. Our method augments the Equivariant GNN framework by integrating an attention mechanism. This attention component allows the model to focus on the most relevant structural features for binding site prediction, significantly enhancing its ability to capture complex spatial patterns and interactions within the protein structure. Our experimental findings underscore the enhanced performance of E(Q)AGNN-PPIS compared to current state-of-the-art approaches, exhibiting gains of 8.33% in the Area Under the Precision-Recall Curve (AUPRC) and 10% in the Matthews Correlation Coefficient (MCC) across benchmark datasets. Additionally, our method demonstrates robust generalization across proteins with varying sequence lengths, outperforming baseline methods.

bioinformatics↗

bioRxiv2024

rpcFold: residual parallel convolutional neural network to decipher RNA folding from RNA sequence

Precise secondary structure information offers deeper insights into the functionality of many RNA molecules. Earlier approaches, which primarily relied on the free-energy minimization, proved inadequate, as RNA often adopts complex folds, especially with increasing length. We present rpcFold, a residual parallel convolutional neural network-based model for RNA secondary structure prediction. From the nucleotide sequences, we compute a base-pairing possibility score using a locally weighted Gaussian function. It captures the intricate canonical and noncanonical pairing patterns, which enhances the modelling of short- and long-range dependencies. The features are mapped into an image that is the input to rpcFold. Our sliding window mechanism accommodates sequences of arbitrary length, and hence explicitly addresses the prediction of pseudoknots, which are often overlooked in prior works. The performance of rpcFold is demonstrated on nested and on non-nested (pseudoknot) base-pairs. While tested on within-family and cross-family benchmark datasets, rpcFold shows improved performance over existing state-of-the-art methods. Additionally, for long nucleotide sequences with complex pseudoknots, rpcFold achieves 71.1% F1-score in jointly predicting nested and non-nested pairs. Performance on unseen RNA families further confirms the robustness and adaptability of our approach. The prediction accuracy of rpcFold, particularly in long-range RNAs and pseudoknots, improves our understanding of RNA functions.

bioinformatics↗

bioRxiv2024

TGFb-ANGPT2-Tie2 axis in cancer-associated fibroblasts reprograms oral cancer cells to embryonic-like cell state with predictive significance of poor prognosis

Myofibroblastic cancer-associated fibroblasts (CAFs) in tumor stroma is identified as poor-prognostic indicator in oral cancer; however, biological mechanisms are largely unexplored. Here, we discovered the role of autocrine or exogenous transforming growth factor beta (TGF{beta}) in inducing Tunica Interna Endothelial cell kinase 2 (Tie2) -signaling through histone deacetylase-mediated downregulation of Tie2-antagonist, Angiopoietin-2 in CAFs, responsible for induction and maintenance of myofibroblastic differentiation. To understand the influence of CAF-specific Tie2-signaling on cancer cell properties, we performed CAF-Cancer cell co-culture and its single-cell RNA sequencing (scRNA-Seq). Distinct clustering of CAFs suggested their transcriptional heterogeneity, driven by TGF{beta}-Tie2 activation. Interestingly, CAF-specific Tie2-signaling was responsible to reprogram cancer cells, producing embryonic-like cell state with increased stemness and EMT signatures. Importantly, both the Tie2-specific gene expression signature as well as reprogrammed cancer cell specific gene expression modules were validated respectively in fibroblasts clusters and malignant cell clusters in two independent earlier reported scRNAseq studies of HNSCC tumors. Highlighting the translatability of our study, the gene expression signature derived from reprogrammed cancer cells showed significant association with poor prognosis in HNSCC patient of TCGA cohort. Pharmacological inhibition of Tie2-signaling in CAFs, significantly abrogated the tumor initiating ability of co-cultured oral cancer cell lines. Overall, combining our molecular and computational analysis, we may propose Tie2 as a novel factor responsible for CAF mediated cancer cell plasticity, associated with aggressive nature of oral cancer. TeaserTie2-signaling is activated in myofibroblasts which impacts the behaviour of malignant cells by inducing cancer cell plasticity to acquire stemness.

cancer biology↗

bioRxiv2024

R-Loop Functions in Brca1-Associated Mammary Tumorigenesis

Excessive R-loops, a DNA-RNA hybrid structure, are associated with genome instability and BRCA1 mutation-related breast cancer. Yet the causality of R-loops in tumorigenesis remains unclear. Here we show that R-loop removal by Rnaseh1 overexpression (Rh1-OE) in Brca1-knockout (BKO) mouse mammary epithelium exacerbates DNA replication stress without affecting homology-directed DNA repair. R-loop removal also diminishes luminal progenitors, the cell of origin for estrogen receptor (ER)-negative BKO tumors. However, R-loop reduction does not dampen spontaneous BKO tumor incidence. Rather, it gives rise to a significant percentage of ER-expressing BKO tumors. Thus, R-loops reshape mammary tumor subtype rather than promoting tumorigenesis.

cancer biology↗

bioRxiv2023

Genetic Variations and Altered Blood mRNA Level of Circadian Genes and BDNF as Risk Factors of Post-Stroke Cognitive Impairment among Eastern Indians

BackgroundPost-stroke cognitive impairment (PSCI) is a clinical outcome in around 30% of post-stroke survivors. BDNF is a major gene in this regard. It regulates and being regulated by circadian rhythm. The circadian genes are correlated with stroke timings at molecular level. However, studies suggesting the role of these on susceptibility to PSCI is limited. AimWe aim here to determine a) genetic risk variants in circadian clock genes, BDNF and b) dysregulation in expression level of CLOCK, BMAL1 and BDNF, that may be associated with PSCI. MethodsBDNF (rs6265G/A, rs56164415C/T), CLOCK (rs1801260T/C, rs4580704G/C) and CRY2 (rs2292912C/G) genes variants were genotyped among 119 post-stroke survivors and 292 controls from Eastern part of India. In addition, we analysed their gene expression in PBMC from 15 PSCI cases and 12 controls. The mRNA data for BDNF was further validated by its plasma level through ELISA. ResultsAmong the studied variants, only rs4580704/CLOCK showed an overall association with PSCI (P = 0.001) and lower BMSE score. Its C allele showed a correlation with attention deficiency. The language and memory impairments showed association with rs6265/BDNF while the CC genotype of rs2292912/CRY2 negatively influenced language and executive function. A significant decrease in gene expression for CLOCK and BDNF in PBMC (influenced by specific genotypes) of PSCI patients was observed than controls. Unlike, Pro-BDNF, plasma level mBDNF was also lower in them. ConclusionsOur results suggest that the circadian genes and BDNF play a role in PSCI on both genetic and transcript level.

neuroscience↗

bioRxiv2023

The molecular prognostic score, a classifier for risk stratification of high-grade serous ovarian cancer

The clinicopathological parameters such as residual tumor, grade, FIGO score are often used to predict the survival of ovarian cancer patients, but the 5-year survival of high grade serous ovarian cancer (HGSOC) still remains around 30%. In recent years, a gene expression based molecular prognostic score (mPS) was developed that showed improved prognosis in several cancers including ovarian cancer. The feature extraction using LASSO-Cox regression was applied on the training data with 10-fold cross validation to obtain 20 predictor genes along with the coefficients to derive mPS. The mPS based prognosis of HGSOC patients was validated using the log-rank test and receiver operator characteristic curve. The AUC of 20 gene-based mPS in predicting the 5-year overall survival was around 0.7 in both the training (n=491) and test datasets (n=491). It was also validated across HGSOC patients (n=7542), data collected from the Ovarian Tumor Tissue Analysis (OTTA) consortium. The mPS showed significant impact (adjusted HR = 6.1, 95% CI of HR= 3.65-10.3; p <0.001) on prognosis of HGSOC. The performance of mPS for the prognosis of survival of HGSOC was substantially better than conventional parameters: FIGO (adjusted HR=1.1, 95% CI=0.97-1.2, p=0.121), residual disease (adjusted HR=1.3, 95% CI= 1.13-1.4, p<0.001), and age (adjusted HR=1.2, 95% CI= 0.98-1.6, p=0.08). It was found that focal-adhesion, Wnt and Notch signaling pathways were significantly (p<0.001) upregulated, whereas antigen processing and presentation (p<0.001) was downregulated in high risk HGSOC cohorts based on mPS stratification. The molecular prognostic score derived from 20-gene signature is found to be the novel robust prognostic marker of HGSOC. It could potentially be harnessed in clinical settings to determine the overall survival of ovarian cancer. The high risk HGSOC patients based on mPS stratification could be benefited from alternative therapies targeting Wnt/ Notch signaling pathways and also immune evasion. Author summaryThe 20-gene signature based molecular prognostic score (mPS) was found to be associated with risk stratification and hence, predicting the overall survival time of HGSOCs. It was applicable in all training HGSOC datasets and across RNA sequencing platforms that also includes the previous reported studies in HGSOC cohorts (Millstein et al., 2020; Talhouk et al., 2020). The 20-gene signature based mPS for the prognosis of overall survival of HGSOC outperformed conventional parameters: age, residual disease and FIGO score. The high or increased risk group of HGSOC based on our mPS stratification was found to have dysregulated pathways of Wnt, Notch, Akt signaling, and antigen presentation. Thus, treatments targeting these pathways might be beneficial for high risk HGSOC and hence anticipated to improve the over-all survival of HGSOC.

genomics↗

bioRxiv2021

ProMoCell and ProModb: web services for analyzing interaction-based functionally localized protein modules in a cell

The modular organization of a cell which can be determined by its interaction network allows us to understand a mesh of cooperation among the functional modules. Therefore, cellular level identification of functional modules aids in understanding the functional and structural characteristics of the biological network of a cell and also assists in determining or comprehending the evolutionary signal. We develop ProFuM-Cell that performs real-time web scraping for generating clusters of the cellular level functional units of an organism. ProFuMCell constructs the Protein Locality Graphs and clusters the cellular level functional units of an organism by utilizing experimentally verified data from various online sources. Also, we develop ProModb, a database service that houses precomputed whole-cell protein-protein interaction network-based functional modules of an organism using ProFuMCell. Our web service is entirely synchronized with the KEGG pathway database and allows users to generate spatially localized protein modules for any organism belonging to the KEGG genome using its real-time web scraping characteristics. Hence, the server will host as many organisms as is maintained by the KEGG database. Our web services provide the users a comprehensive and integrated tool for an efficient browsing and extraction of the spatial locality-based protein locality graph and the functional modules constructed by gathering experimental data from several interaction databases and pathway maps. We believe that our web services will be beneficial in pharmacological research, where a novel research domain called modular pharmacology has initiated the study on the diagnosis, prevention, and treatment of deadly diseases using functional modules. Graphical TOC Entry O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=111 SRC="FIGDIR/small/446966v2_ufig1.gif" ALT="Figure 1"> View larger version (21K): org.highwire.dtl.DTLVardef@145374borg.highwire.dtl.DTLVardef@13121aeorg.highwire.dtl.DTLVardef@9a3681org.highwire.dtl.DTLVardef@707985_HPS_FORMAT_FIGEXP M_FIG C_FIG

bioinformatics↗

bioRxiv2021

Protein Interaction Network-based Deep Learning Framework for Identifying Disease-Associated Human Proteins

Infectious diseases in humans appear to be one of the most primary public health issues. Identification of novel disease-associated proteins will furnish an efficient recognition of the novel therapeutic targets. Here, we develop a Graph Convolutional Network (GCN)-based model called PINDeL to identify the disease-associated host proteins by integrating the human Protein Locality Graph and its corresponding topological features. Because of the amalgamation of GCN with the protein interaction network, PINDeL achieves the highest accuracy of 83.45% while AUROC and AUPRC values are 0.90 and 0.88, respectively. With high accuracy, recall, F1-score, specificity, AUROC, and AUPRC, PINDeL outperforms other existing machine-learning and deep-learning techniques for disease gene/protein identification in humans. Application of PINDeL on an independent dataset of 24320 proteins, which are not used for training, validation, or testing purposes, predicts 6448 new disease-protein associations of which we verify 3196 disease-proteins through experimental evidence like disease ontology, Gene Ontology, and KEGG pathway enrichment analyses. Our investigation informs that experimentally-verified 748 proteins are indeed responsible for pathogen-host protein interactions of which 22 disease-proteins share their association with multiple diseases such as cancer, aging, chem-dependency, pharmacogenomics, normal variation, infection, and immune-related diseases. This unique Graph Convolution Network-based prediction model is of utmost use in large-scale disease-protein association prediction and hence, will provide crucial insights on disease pathogenesis and will further aid in developing novel therapeutics.

bioinformatics↗

bioRxiv2021

Deciphering biological evolution exploiting the topology of Protein Locality Graph

The conventional sequence comparison-based evolutionary studies ignore other evolutionary constraints like interaction among proteins, functions of proteins and genes etc. A lot of speculations exist in literature regarding the presence of species divergence at the level of the Protein Interaction Networks. Additionally, it has been conjectured that the intra-module connections stay conserved whereas the inter-module connections change during evolution. The most important components of the biological networks are the functional modules which are more conserved among the evolutionary closer species. Here, we demonstrate an alternative method to decipher biological evolution by exploiting the topology of a spatially localized Protein Interaction Network called Protein Locality Graph (PLG). Our lossless graph compression from PLG to a power graph called Protein Cluster Interaction Network (PCIN) results in a 90% size reduction and aids in improving computational time. Further, we exploit the topology of PCIN and demonstrate our capability of deriving the correct species tree by focusing on the cross-talk between the protein modules exclusively. Our results provide new evidence that traces of evolution are not only present at the level of the Protein-Protein Interactions, but are also very much present at the level of the inter-module interactions.

bioinformatics↗