Pangenome Graph-Based HIV Drug Resistance Profiling Pipeline

A Snakemake-based workflow for detecting HIV drug resistance mutations (DRMs) from NGS data. The pipeline supports three alignment strategies:

• Bowtie2 – conventional linear-reference alignment
• LAST – adaptive-scoring alignment (with parameters trained via LAST-TRAIN)
• VG Giraffe / VG Map – graph-based alignment against a pangenome constructed from diverse HIV-1 group M reference genomes

The workflow uses the VG toolkit, MAFFT, and standard NGS tools to construct a pangenome graph (when selected), align reads, call variants, convert results to AAVF, and optionally generate HIV drug resistance interpretations using HIVdb.

What This Pipeline Does

This pipeline performs the following key steps:

1. Input Preparation

• Accepts curated HIV reference genomes (FASTA)
• Accepts raw NGS reads (FASTQ)

2. Graph Construction (Only when using VG Alignment)

If the VG-based alignment strategy is selected, the pipeline will:

• Perform multiple sequence alignment using MAFFT
• Construct a variation graph using vg construct
• Index the graph using vg index, vg gbwt, and related VG indexing tools

(This entire step is skipped for Bowtie2 and LAST.)

3. Alignment

All supported alignment methods are unified here:

• Bowtie2 — standard linear alignment to HXB2 or a user-specified reference
• LAST — adaptive-scoring alignment trained with LAST-TRAIN, improving mapping accuracy for divergent sequences
• VG Giraffe — graph-based alignment to the constructed pangenome graph

Each method outputs SAM/BAM alignment files for downstream processing.

4. Variant Calling

• Calls variants from either linear or graph alignments using vg call
• Produces per-sample VCF files
• Converts VCF → AAVF using a custom Python script

5. Drug Resistance Mutation Identification & Interpretation

• Queries AAVF files against HIVdb using the sierrapy client
• Produces a structured HIV drug resistance profile

Installation

This pipeline requires:

• Miniconda
• Snakemake

All tool dependencies are managed via Snakemake’s --use-conda functionality.

Step 1: Install Miniconda

Follow the official installation guide: https://docs.conda.io/en/latest/miniconda.html

Verify the installation:

conda list

Step 2: Install Snakemake

We recommend using mamba for faster environment resolution:

conda install -c conda-forge mamba
mamba create -c conda-forge -c bioconda -n <env_name> snakemake
conda activate <env_name>

Test the installation:

snakemake --help

Download the Pipeline

git clone https://github.com/bioinfodlsu/hiv_hts_pipeline.git
cd hiv_hts_pipeline

Quick Start Guide

Dry Run

snakemake --use-conda --cores N -np

Run the Pipeline

snakemake --use-conda --cores N -p

Visualize the DAG

snakemake --use-conda --cores all --dag | dot -Tpdf > dag.pdf

Running the Pipeline on Your Own Data

1. Edit config/config.yaml

Set:

• Your email (for NCBI sequence downloads, if needed)
• Paths to your reference FASTA and FASTQ files
• Output directory
• Alignment method (vg_giraffe, bowtie2, last)
• Optional: use_clustered_refs: true to cluster references before graph construction

Ensure paths are correct and unused samples/references are commented out.

Contact

For issues, questions, or contributions, open an issue: https://github.com/bioinfodlsu/hiv_hts_pipeline/issues