Skip to content

Latest commit

 

History

History
200 lines (133 loc) · 8.11 KB

README.md

File metadata and controls

200 lines (133 loc) · 8.11 KB
Raw

mPies: metaProteomics in environmental sciences DOI

mPies is a workflow to create annotated databases for metaproteomic analysis.

This workflow uses three different databases for a metagenome (i) OTU-table, (ii) assembled-derived, (iii) and unassembled-derived to build a consensus of these databases and increase the mapping sensitivity.

If you use mPies for your research, please cite our publication:

Werner, J., Géron, A., Kerssemakers, J. et al. mPies: a novel metaproteomics tool for the creation of relevant protein databases and automatized protein annotation. Biol Direct 14, 21 (2019) doi: 10.1186/s13062-019-0253-x

Installation

The easiest way is to use bioconda and create a new environment. For a faster installation, consider installing micromamba and replace conda with mamba in all runs.

conda env create -n mpies --file conda_env.yml
conda activate mpies

Usage

mPies consists of two parts: database creation and annotation. Both parts are written in Snakemake.

# database creation
snakemake --snakefile database_creation.smk --configfile database_creation.json --cores 28

# annotation
snakemake --snakefile annotation.smk --configfile annotation.json --cores 28

Detailed explanation of the mpies workflow

Database creation

Preprocessing

The preprocessing trims the raw reads and combines the single reads into one file.

Amplicon-derived proteome file

In order to create the amplicon-derived proteome file, there are two possibilities. If amplicon data is available, then a text file with the taxon names (one per line) is used for downloading the proteomes from UniProt. If no amplicon data is available, you can set the option config["otu_table"]["run_singlem"] to true and a taxon file is created with SingleM (this tool detects OTU abundances based on metagenome shotgun sequencing data).

Important Note: SingleM

SingleM currently cannot be used as it uses orator as a dependency that still relies on Python 3.5. As long as this is not fixed (and since the last code edit in the orator Github repository is 3 years ago (state of this comment: September, 2022), it is not likely that this will happen anytime soon), SingleM cannot be used and is deactivated until further notice.

Functional-derived subset

It is also possible to create a subset derived from UniProt based not only on taxonomy but to also restrict the gene and functional names instead of downloading the entire proteomes for the taxa of interest. To do so, a TOML file should be created (see example below)

Taxonomy = [
    "Bacteria"
]
Gene_names = [
     "dnaK",
     "soxA"
]
Protein_names = [
    "Heat shock protein 70", # something commented
]

and the path needs to be set in the snakemake configuration (config["functional_subset"]["toml_file"]).

Assembled-derived proteome file

If only raw data is available, it is possible to run an assembly with MEGAHIT or metaSPAdes (set config["assembled"]["run_assembly"] to true and config["assembled"]["assembler"] to megahit or metaspades). Please keep in mind that assemblies can take a lot of time depending on the size of the dataset. If you already have an assembly, set config["assembled"]["run_assembly"] to false and create a symlink of your assembly into {sample}/assembly/contigs.fa. If you have no gene calling yet, remember to set config["assembled"]["run_genecalling"] to true.

If you have both assembly and gene calling already performed, set config["assembled"]["run_assembly"] and config["assembled"]["run_genecalling"] to false and create a symlink of the assembled proteome into {sample}/proteome/assembled.faa.

Unassembled-derived proteome file

To create the unassembled-derived proteome file, FragGeneScan is used (and prior to that a fastq-to-fasta conversion).

Postprocessing

During the postprocessing, the all three proteomes are combined into one file. Short sequences (< 30 amino acids) are deleted and all duplicates are removed. Afterwards, the fasta headers are hashed to shorten the headers (and save some disk space).

Annotation

Preprocessing

For now, the identified proteins are inferred from ProteinPilot. The resulting Excel file is used to create a protein fasta file that only contains the identified proteins. Taxonomic and functional analysis are conducted for the identified proteins.

Taxonomical annotation

The taxonomic analysis is performed with blast2lca from the MEGAN package. Per default, the taxonomic analysis is set to false in the snake config file.

Some prerequisites are necessary to run the taxonomic analysis for the created proteome fasta file.

  1. Download and unzip the file prot_acc2tax-June2018X1.abin.zip for MEGAN.

  2. Download the nr.gz fasta file from NCBI (size: 40 GB).

wget ftp://ftp.ncbi.nlm.nih.gov/blast/db/FASTA/nr.gz
wget ftp://ftp.ncbi.nlm.nih.gov/blast/db/FASTA/nr.gz.md5
md5sum -c nr.gz.md5

If the checksum does not match, the download was probably not complete. wget -c continues a partial download.

  1. Create a diamond database of the file nr.gz.
diamond makedb --threads <number_of_threads> --in nr.gz --db nr.dmnd
  1. Now you can set config["taxonomy"]["run_taxonomy"] to true and run snakemake. Remember to set the paths for the diamond database, the binary of blast2lca and the path to the file prot_acc2tax-Jun2018X1.abin. Please note that diamond blastp takes a very long time to execute.
Functional annotation

Different databases can be used to add functional annotation. Per default, the funtional annotation is set to false.

COG

In order to use the COG database, some prerequisites have to be fulfilled before.

  1. Download the necessary files from the FTP server.
wget ftp://ftp.ncbi.nih.gov/pub/COG/COG2014/data/prot2003-2014.fa.gz
wget ftp://ftp.ncbi.nih.gov/pub/COG/COG2014/data/cog2003-2014.csv
wget ftp://ftp.ncbi.nih.gov/pub/COG/COG2014/data/cognames2003-2014.tab
wget ftp://ftp.ncbi.nih.gov/pub/COG/COG2014/data/fun2003-2014.tab
  1. Create a diamond database of the file prot2003-2014.fa.gz.
diamond makedb --threads <number_of_threads> --in prot2003-2014.fa.gz --db cog.dmnd
  1. Now you can set config["functions"]["run_cog"]["run_functions_cog"] to true and run snakemake. Remember to set the paths for the diamond database and the files cog_table, cog_names, and cog_functions.
UniProt/GO

In order to use the GO ontologies included in the UniProt database (SwissProt or TrEMBL), some prerequisites have to be fulfilled before.

  1. Download the necessary files from the FTP server.
# SwissProt
wget ftp://ftp.uniprot.org/pub/databases/uniprot/current_release/knowledgebase/complete/uniprot_sprot.fasta.gz
wget ftp://ftp.uniprot.org/pub/databases/uniprot/current_release/knowledgebase/complete/uniprot_sprot.dat.gz

# TrEMBL
wget ftp://ftp.uniprot.org/pub/databases/uniprot/current_release/knowledgebase/complete/uniprot_trembl.fasta.gz
wget ftp://ftp.uniprot.org/pub/databases/uniprot/current_release/knowledgebase/complete/uniprot_trembl.dat.gz

Please note that TrEMBL is quite large (29 GB for uniprot_trembl.fasta.gz and 78 GB for uniprot_trembl.dat.gz).

  1. Create a diamond database of the fasta file (here the SwissProt database will be used)
diamond makedb --threads <number_of_threads> --in uniprot_sprot.fasta.gz --db sprot.dmnd
  1. Use the dat file downloaded from UniProt to create a table with protein accessions and GO annotations
./main.py prepare_uniprot_files -u .../uniprot_sprot.dat.gz -t .../sprot.table.gz

Please note that input and output files must be/are compressed with gzip.

  1. Now you can set config["functions"]["run_uniprot"]["run_functions_uniprot"] to true and run snakemake.

Test data

The test data set is a subset from the Ocean Sampling Day (first 18,000 lines for each read file), Accession number ERR770958 obtained from https://www.ebi.ac.uk/ena/data/view/ERR770958). The data is deposited in the test_data directory of this repository.