NGS Analysis Pipelines

This repository contains bioinformatics workflows and analysis pipelines developed using Python and R with reference to my M.Sc. Biotechnology & Bioinformatics coursework and academic projects at the Institute of Bioinformatics and Applied Biotechnology (IBAB), Bengaluru. The repository includes reproducible pipelines for genomics, transcriptomics, epigenomics, metagenomics, and sequence analysis using Python, R, and widely used bioinformatics tools.

Contents

Folder	Analysis	Language	Key Tools
01_Microarray_Analysis	Differential expression from Affymetrix microarray data (E-MTAB-6095: RNase I-deficient E. coli vs WT)	R	oligo, limma
02_RNAseq_Analysis	(1) Alignment, (2) RPKM quantification and DE analysis	Python + R	Bowtie2 · Rsubread, edgeR
03_Alternative_Splicing	(1) Splice-aware alignment, (2) splice graph/junction analysis	Python + R	HISAT2 · SGSeq
04_Genome_Assembly	De novo genome assembly, QC, gene prediction, scaffolding	Python	Velvet, QUAST, Augustus, RagTag, BUSCO
05_Variant_Calling	SNP/Indel calling pipeline following GATK best practices	Python	BWA-MEM, Picard, GATK
06_ChIP_Seq	Peak calling, annotation, and motif discovery	Python	Bowtie2, MACS2, HOMER, BEDTools, MEME
07_QIIME2_Metagenomics	16S amplicon microbial community & diversity analysis	Python	QIIME 2
08_Bisulfite_Sequencing	DNA methylation mapping at single-base resolution	Python	Bismark, Bowtie2, samtools

Why Python for some pipelines and R for others?

Tools like Bowtie2, samtools, GATK, MACS2, HOMER, bedtools, MEME, QIIME2, and Bismark are standalone command-line programs, there's no R or Python package that reimplements them, so any script that runs them is just calling out to the same binaries. Rather than raw shell scripts, those pipelines are written in Python, using the subprocess module to call each tool, this keeps the orchestration logic readable, testable, and easy to extend (e.g. adding loops, config files, logging) compared to bash.

The statistics, heavy steps (differential expression, splice graph analysis) use Bioconductor packages: limma, edgeR, SGSeq, that only exist in R. There's no equivalent Python library that reproduces the same statistical methods, so those stay in R rather than being awkwardly reimplemented.

Usage

Each script/folder assumes:

• Raw input files (FASTQ/CEL/BAM, reference genome, annotation) are placed in the working directory or a data/ subfolder as indicated in the script.
• Required command-line tools are installed and on $PATH.
• For R scripts, the listed Bioconductor/CRAN packages are installed.

Edit the variables at the top of each script (sample names, file paths, reference genome) before running.

# Example: run the ChIP-Seq pipeline
cd 06_ChIP_Seq
python3 chipseq_pipeline.py

# Example: run the microarray DE analysis in R
setwd("01_Microarray_Analysis")
source("microarray_de_analysis.R")

Notes

• Scripts are adapted from analyses originally run on lab/course datasets; paths have been generalized for portability. Substitute your own reference genome, annotation, and sample FASTQ/CEL files as needed.
• Each pipeline was documented step-by-step to support learning and reproducibility, not just execution — comments explain the purpose of each step (QC, alignment, normalization, statistical testing, visualization).