Skip to content

Latest commit

 

History

History
148 lines (112 loc) · 6.29 KB

README.md

File metadata and controls

148 lines (112 loc) · 6.29 KB

Snakemake GitHub actions status

Snakemake workflow: mg_assembly

A Snakemake workflow for Genome Resolved Metagenomics

Features

  • Preprocessing:
    • FASTQ processing with fastp.
    • Mapping of preprocessed reads against the host(s) and possible contaminants with bowtie2. Skip if no host is provided.
      • Useful for environmental genomics (there is no host in soil, duh!), or
      • environments that can have multiple genomes, like in mycorrhiza, where plant, fungus and even insects, can be there.
    • Assembly-free statistics with kraken2, nonpareil and singlem.
  • Assembly of non-host reads with megahit.
    • Coassembly strategies denoted in the samples.tsv file. See below.
    • Taxonomic annotation of contigs with kraken2
    • Assembly quantification with bowtie2 and coverm
  • Bacterial metagenomics:
    • Binning with CONCOCT, Maxbin2, MetaBAT2, and aggregated with MAGScoT.
    • Annotation with quast (mag and contig lengths), gtdbtk (taxonomy), dram (functions) and checkm2 (completeness and contamination)
    • Dereplication with dRep, using multiple secondary ANIs, in case you need one for read mapping (eg. 95%), and a different for something like pangenomics (98 and 99%).
    • Quantification with bowtie2 and coverm. One per secondary ANI
  • Viral metagenomics:
    • Identification and clustering with genomad, bbmap and mmseqs.
    • Quantification with bowtie2 and coverm.
    • Annotation with dram, virsorter2, checkv and quast.
  • Module reporting with multiqc, assisted with samtools and fastqc.

Usage

  1. Make sure you have conda, mamba and snakemake installed.

    conda --version
snakemake --version
mamba --version

  2. Clone the git repository in your terminal and get in:

    git clone [email protected]:3d-omics/mg_assembly.git
cd mg_assembly

  3. Test your installation by running the test data. It will download all the necesary software through conda / mamba. It should take less than 5 minutes.

    snakemake --use-conda --cores 8 test

  4. Run it with your own data:

    1. Edit config/samples.tsv and add your samples names, a library identifier to differentiate them, where are they located, the adapters used, and the coassemblies each sample will belong to.
    sample_id	library_id	forward_filename	reverse_filename	forward_adapter	reverse_adapter	assembly_ids
sample1	lib1	resources/reads/sample1_1.fq.gz	resources/reads/sample1_2.fq.gz	AGATCGGAAGAGCACACGTCTGAACTCCAGTCA	AGATCGGAAGAGCGTCGTGTAGGGAAAGAGTGT	sample, all
sample2	lib1	resources/reads/sample2_1.fq.gz	resources/reads/sample2_2.fq.gz	AGATCGGAAGAGCACACGTCTGAACTCCAGTCA	AGATCGGAAGAGCGTCGTGTAGGGAAAGAGTGT	all

    In the assembly_ids you can name all the coassemblies each library will belong to. If you don't want to use a sample (a blank or a failed sample), leave the field empty.

    1. Edit config/features.yml with reference databases:
    hosts:  # Add more in case of multi-host, remove entries in case of environmental sample
  chicken: resources/reference/chicken_39_sub.fa.gz
  # pig: resources/reference/pig.fa.gz

magscot:
  pfam_hmm: workflow/scripts/MAGScoT/hmm/gtdbtk_rel207_Pfam-A.hmm.gz
  tigr_hmm: workflow/scripts/MAGScoT/hmm/gtdbtk_rel207_tigrfam.hmm.gz

databases:  # The pipeline does not provide or generate them. There are scripts tho.
  checkm2: resources/databases/checkm2/20210323/uniref100.KO.1.dmnd
  checkv: resources/databases/checkv/20230320/checkv-db-v1.5/
  dram: resources/databases/dram/20230811
  genomad: resources/databases/genomad/genomad_db_v1.7
  gtdbtk: resources/databases/gtdbtk/release214
  kraken2:  # add entries as necessary
    refseq500: resources/databases/kraken2/kraken2_RefSeqV205_Complete_500GB/20220505/
  singlem: resources/databases/singlem/S3.2.1.GTDB_r214.metapackage_20231006.smpkg.zb
  virsorter2: resources/databases/virsorter2/20200511/
    1. Edit config/params.yml with execution parameters. The defaults are reasonable.

  5. Run the pipeline

    # make sure firsthand that you have all the databases above properly installed
snakemake --use-conda --cores 8  # locally
snakemake --use-conda --cores 24 --jobs 100 --executor slurm  # in slurm

  6. Output:

    The main outputs are:

    1. results/prokaryotes/annotate/: MAG annotations.
    2. results/prokaryotes/quantify/: MAG and contig-wise quantifications.
    3. There is an experimental pipeline for viral identification with a similar structure. See below. The results are in results/viruses/.
    4. MultiQC html reports and tables next to the main modules: preprocess, assemble, prokaryotes and viruses.

Rulegraph

rulegraph_simple

References