Shotgun Metagnomics Pipeline

Description

This is a Nextflow re-implementation of the original pipeline used by Computational and Systems Biology Group 5 (CSB5) at the Genome Institute of Singapore (GIS).

中文文档

Development plan

• [ ] Add customized HUMAnN2 to a conda channel
• [ ] Add nf-core style documentation
  • [x] Output description
  • [ ] Installation
  • [ ] Usage
  • [ ] Reference databases

Features

• The new DSL2 syntax for pipeline modularity and reusabiligy
• Dockerfile for each software (all containers can be found at DockerHub)
• Conda recipe for each software/step
• Configuration for local execution (server), GIS HPC (using SGE schedular), AWS batch and AWS auto-scaling cluster

Dependencies

Main pipeline

• Nextflow
• Java Runtime Environment >= 1.8

Quality control and host DNA decontamination

• Fastp (>=0.20.0): Adapter trimming, low quality base trimming
• BWA (>=0.7.17): Host DNA removal
• Samtools (>=1.7): Host DNA removal

Reference based analysis

• Kraken2 (>=2.0.8-beta) + Bracken (>=2.5): Taxonomic profiling
• MetaPhlAn2 (>=2.7.7): Taxonomic profiling
• HUMAnN2 (>=2.8.1): Pathway analysis. The following two files are modified to read a SAM file from standard input (See Running HUMAnN2 with reduced disk storage):
  • humann2.py
  • search/nucleotide.py
• SRST2 (=0.2.0): Resistome profiling

Setup and configuration

• Installation instruction

Usage

• Pipeline parameters
• Run on GIS HPC cluster (SGE scheduler)
• Run on AWS configured for CSB5
• Run on NSCC

Run with docker

$ shotgunmetagenomics-nf/main.nf -profile docker --read_path PATH_TO_READS

Run on AWS batch (AWS batch configuration tutorial)

• IAM configuration (set environment variables AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY, AWS_DEFAULT_REGION)
• Batch compute environment & job queue
• Customized AMI (AWS ECS optimized linux + awscli installed with miniconda)

$ shotgunmetagenomics-nf/main.nf -profile awsbatch --awsqueue AWSBATCH_QUEUE --awsregion AWS_REGION --bucket-dir S3_BUCKET --outdir S3_BUCKET 

You can specifiy multiple profiles separated by comma, e.g. -profile docker,test.

Run multiple profilers

$ shotgunmetagenomics-nf/main.nf -profile gis --profilers kraken2,metaphlan2 --read_path PATH_TO_READS

Usage cases

• Chng et al. Whole metagenome profiling reveals skin microbiome dependent susceptibility to atopic dermatitis flares. Nature Microbiology (2016)
• Nandi et al. Gut microbiome recovery after antibiotic usage is mediated by specific bacterial species. BioRxiv (2018)
• Chng et al. Cartography of opportunistic pathogens and antibiotic resistance genes in a tertiary hospital environment. BioRxiv (2019)

Adding a module

1. Write a module and put it into modules/
2. Add to the main script main.nf
3. Modify the configuration file conf/base.config to add resources required (for GIS users, modify conf/gis.config as well for the specific conda envrionment)
4. Add conda and docker files for the new module

Contact

Chenhao Li: lich@gis.a-star.edu.sg, lichenhao.sg@gmail.com