Pan-Genome and Comparative Genomics of Elizabethkingia anophelis

Project Overview

This repository contains a hands-on comparative genomics and pan-genome analysis of multiple Elizabethkingia anophelis strains. The project was built end-to-end using Linux-based bioinformatics tools and Python to explore core vs accessory genome structure, strain-level gene content variation, and evolutionary patterns driven by horizontal gene transfer (HGT).

Rather than focusing on pre-built pipelines or GUI tools, this project emphasizes building and debugging real-world genomics workflows, feature engineering, and biologically grounded exploratory analysis.

---

Scientific Goals

• Characterize the core and accessory genome of E. anophelis
• Identify strain-level gene content divergence
• Detect clonal groups and divergent lineages
• Explore horizontal gene transfer through accessory genome structure
• Prepare a foundation for AMR gene integration and phylogenomics

---

Dataset

Strains analyzed:

• R26
• JUNP353
• Eli4
• Eli5
• Eli6
• F3201
• ES-CHC077
• 296-96

All genomes were obtained from NCBI RefSeq/GenBank and processed locally.

---

Methods & Pipeline

1. Genome Annotation

Tool: Prokka

• Annotated each genome independently
• Generated standardized GFF, FAA, FFN, and annotation summaries
• Extracted genome-level features for comparative analysis

Key outputs:

• prokka_<strain>/*.gff
• prokka_<strain>/*.tsv
• prokka_<strain>/*.txt

---

2. Baseline Comparative Features

From Prokka outputs, the following features were extracted:

• Genome size (Bases)
• CDS count
• rRNA and tRNA counts
• Contig count
• Efflux-related gene counts

These were merged into a master feature table and used for initial exploratory PCA.

Purpose:

• Sanity-check strain-level structure
• Establish baseline genomic differences

---

3. Pan-Genome Analysis

Tool: Roary

All Prokka GFF files were used as input to build the pan-genome.

Command (simplified):

roary -e -n -v -p 4 -f roary_out roary_input/*.gff

Key outputs:

• gene_presence_absence.csv
• gene_presence_absence.Rtab
• core_gene_alignment.aln
• summary_statistics.txt

---

Pan-Genome Summary

From Roary:

• Core genes: 2,798
• Shell genes: 2,131
• Cloud genes: 1,871
• Total pan-genome size: 6,800 genes

Interpretation:

• ~41% core genome
• ~60% accessory genome (shell + cloud)

This indicates high genomic plasticity and extensive strain-specific gene content, consistent with horizontal gene transfer and niche adaptation.

---

Accessory Genome PCA (Key Result)

PCA was performed on the accessory genome gene presence/absence matrix (thousands of genes) to explore lineage structure and HGT-driven variation.

Key findings:

• Eli4 / Eli5 / Eli6 form a tight cluster → near-clonal gene content
• JUNP353 is a strong outlier → highly divergent accessory genome
• ES-CHC077 & 296-96 form a distinct sub-cluster
• R26 and F3201 show intermediate divergence

This demonstrates clear strain-level evolutionary structure driven by accessory genome variation.

---

Biological Interpretation

The large accessory genome fraction suggests that E. anophelis exhibits substantial genomic flexibility. Divergence patterns indicate:

• Strong clonality within specific strain groups
• Extensive strain-specific gene acquisition
• Likely horizontal gene transfer shaping lineage structure

These patterns are consistent with opportunistic pathogens and environmentally adaptable bacteria.

---

Tools & Technologies

• Prokka (genome annotation)
• Roary (pan-genome analysis)
• Python (pandas, scikit-learn, matplotlib)
• Linux command-line workflows
• Conda environments

---

Reproducibility

All analyses were performed using command-line tools and Python notebooks. Environment-specific issues (Perl and Conda dependencies) were resolved manually, reflecting real-world genomics pipeline development.

---

Current Status

Completed:

• Genome annotation
• Pan-genome construction
• Accessory genome PCA
• Lineage-level interpretation

---

Planned Extensions

• Identification of strain-specific genes (e.g., JUNP353 unique genes)
• Integration of AMR gene prediction (AMRFinderPlus)
• Core genome phylogenetic tree reconstruction
• Mapping AMR genes onto pan-genome categories
• Accessory genome size vs resistance profiling

---

Motivation

This project is part of a learning-by-building approach to bioinformatics, focusing on real datasets, real tools, and real biological questions rather than tutorial-driven workflows.

---

Author

Nischay Mehta
B.Tech Bioinformatics Aspirant (Class of 2027)
Focus: Comparative genomics, microbial genomics, and ML-driven exploratory analysis

---

Disclaimer

This is an educational and exploratory research project. Results are intended for learning and portfolio demonstration and are not clinical or diagnostic.