An end-to-end, reproducible computational framework for transcriptomic biomarker discovery and validation.
This repository demonstrates how raw RNA-seq or microarray data can be processed, analyzed, and modeled using a combination of statistical and machine learning methods.

• Gained foundational understanding of transcriptomic technologies (RNA-seq, microarray) and primary data repositories including GEO, SRA, NCBI, and ArrayExpress.
• Analyzed raw RNA-seq data from FASTQ to quantified counts using splice-aware aligners (STAR, HISAT2) and quasi-mappers (Salmon, Kallisto) for robust gene expression profiling.
• Retrieved and processed GEO datasets using GEOquery, followed by manual metadata curation for comparative transcriptomic analysis.
• Performed differential gene expression analysis using edgeR, limma, and DESeq2, including normalization, outlier detection, and batch effect correction.
• Applied downstream visualization and statistical methods: heatmaps, boxplots, PCA, UMAP, t-SNE, hierarchical clustering, volcano plots, and phylogenetic analyses.
• Conducted functional enrichment analysis using DAVID, STRING, Enrichr, Reactome, and GSEA to identify enriched pathways and biological processes associated with Mpox virus infection.
• Predicted candidate therapeutic compounds using DSigDB based on DEG signatures.
• Validated key DEGs as potential biomarkers via ROC curve analysis (AUC-based).
• Developed and benchmarked machine learning models (Random Forest, Extra Trees, SVM) to identify predictive biomarkers, including model tuning and feature importance ranking.
• Utilized a range of bioinformatics tools and libraries:
  R (DESeq2, limma, edgeR, GOplot, ggplot2, pROC),
  Cytoscape (ClueGO, CluePedia, cytoHubba, MCODE),
  Python (PyCaret, scikit-learn, pandas), and TBtools.
• First-author publication:
  Debnath, J.P., et al. (2025). Identification of potential biomarkers for 2022 Mpox virus infection: a transcriptomic network analysis and machine learning approach. Scientific Reports, 15, 2922.

1️⃣ Preprocessing & Quality Control    →  FastQC, Trimmomatic  
2️⃣ Alignment & Quantification         →  STAR / Salmon / featureCounts  
3️⃣ Differential Expression Analysis   →  DESeq2, edgeR, limma  
4️⃣ Functional Enrichment Analysis     →  GSEA, Enrichr, GOstats  
5️⃣ Network Modeling                   →  Identify hub genes, functional modules  
6️⃣ Machine Learning Benchmarking      →  PyCaret (AUPRC, AUROC, F1, MCC)  

Raw RNA-seq and microarray datasets used in this pipeline can be downloaded from public repositories:

• SRA (Sequence Read Archive)
• GEO (Gene Expression Omnibus)

This project is distributed under the MIT License.
See LICENSE for details.