Marine Omics Annotation Pipeline

Performs functional annotation on mRNA and protein sequences. This includes predicted transcript (mRNA/CDS) and protein sequences from gene models (eg genome obtained via NCBI or gene models generated in house). It also includes assembled transcripts and corresponding predicted proteins from a denovo transcriptome assembly.

Note that moat primarily performs annotation on proteins. If provided, nucleotide sequences (transcript sequences) are used only for blastx searches to identify hits that might otherwise be obscured by low quality protein prediction (eg due to a frameshift). No attempt is made to annotate non-coding transcripts.

graph TD;
	prot-->signalp;
	prot-->tmhmm;
	prot-->blastp;
	cds-->blastx;
	swissprot-->blastp;
	swissprot-->blastx;
	prot-->interproscan;
	signalp-->R;
	tmhmm-->R;
	blastp-->R;
	blastx-->R;
	interproscan-->R;
	R-->annotation.tsv

Loading

Installation

First install and configure nextflow. See here for instructions specific to JCU machines (zodiac, genomics1, genomics2)

Quick Start

Assuming your predicted proteins are in protein.fasta, run moat as follows on the JCU HPC (zodiac).

nextflow run marine-omics/moat -latest -profile zodiac -r main --prot protein.fasta

If desired you may also provide a set of transcripts (nucleotide sequences for your proteins). These should have identical IDs to your proteins. Assuming they are in a file called transcripts.fasta you would run moat as follows;

nextflow run marine-omics/moat -latest -profile zodiac -r main --cds transcripts.fasta --prot protein.fasta

Outputs

After running moat you should find raw outputs from each of the individual annotation tools (blast, interproscan, hmmscan etc) in the output directory. There should also be a file, final_table/annotation.tsv in which all of these results are joined into a single table. This table has columns;

• id: Your original protein ID
• evalue: E value of the best blast hit
• Swissprot_acc: Swissprot accession of the best blast hit
• method: Which blast method (blastp, blastx) gave the best blast hit
• ipr_go: GO terms assigned by Interproscan based on the presence of conserved domains and protein family membership
• Pfam_desc: Names of Pfam domains identified by Interproscan
• Pfam_acc: Accessions of Pfam domains identified by Interproscan
• SUPERFAMILY_desc: Name of the superfamily assigned by Interproscan
• SUPERFAMILY_acc: Accession of the superfamily assigned by Interproscan
• signal_peptide: Whether a signal peptide was identified by SignalP
• tmhmm_topology: Transmbrane domain Topology identified by TMHMM. See here for an explanation of this format

Obtaining more annotations

The Swissprot_acc field in the moat output provides a key for obtaining many more detailed annotations. These can be looked up from the uniprot website and easily joined to your table.

First extract a list of all swissprot accessions for upload

cat annotation.tsv | grep -v 'Swissprot' | awk '{print $3}' | sort -u > swissprot_acc.txt

Upload these accessions to https://www.uniprot.org/id-mapping.

When your mapping job is complete you will be presented with a list of swissprot entries corresponding to your identifiers. The default is to show just columns with basic information, however, you can add many more columns of information to the table by clicking "Customize columns". In particular you will probably want to add a column for Gene Ontology IDs.

Once you have finished adding columns you can download the full table of results. Choose excel format for download. You can then join this information with your moat table like this;

library(tidyverse)

moat_annotations <- read_tsv("outdir/final_table/annotation.tsv")

swissprot_annotations <- readxl::read_excel("uniprot-download.xlsx")

moat_annotations %>% left_join(swissprot_annotations,by=c("Swissprot_acc"="From"))

Databases

On JCU systems moat should already be configured with access to the databases it needs. These include;

• The Uniprot Swissprot database of annotated proteins (for blastp and blastx processes)
• A collection of databases required for Interproscan

If these databases are not available on your system you will need to install them as follows;

Swissprot

Download and build using commands below.

wget https://ftp.uniprot.org/pub/databases/uniprot/current_release/knowledgebase/complete/uniprot_sprot.fasta.gz
makeblastdb -in uniprot_sprot.fasta -dbtype 'prot' -parse_seqids -out swissprot

This will produce a set of files with names matching swissprot.*. You can tell moat to find these files by adding the following setting to your local.config

params.blastdb="<path/to/folder>/swissprot"

Interproscan

Instructions below assume you are installing Interproscan data in the directory /scratch/ipr/data. Adjust according to your preferred installation path.

IPR=5
IPRSCAN=5.59-91.0

cd /scratch/ipr

wget -O interproscan-data-$IPRSCAN.tar.gz ftp://ftp.ebi.ac.uk/pub/software/unix/iprscan/$IPR/$IPRSCAN/alt/interproscan-data-$IPRSCAN.tar.gz
tar -xvzf interproscan-data-$IPRSCAN.tar.gz

Once unpacked this data directory needs to be mounted into the container at a specific path. We do this by configuring runOptions for singularity / docker as follows;

singularity.runOptions = '--bind /scratch/ira/ipr/interproscan/data:/opt/interproscan/data'