IMG Omics Annotation WDL

Setup

Requirements

• Cromwell (and associated Java dependencies)
• Input binaries, databases, etc. specified in the inputs.json file

Docker image

This pipeline uses a docker image. You can see the dockerfile and latest tag(version) here:
https://hub.docker.com/repository/docker/bfoster1/img-omics

Input

Input options are listed in the inputs.json file and include locations of input databases, locations of necessary executables, and Boolean variables to toggle run options on and off.

Currently, the inputs have some executables' paths written in using the absolute system paths on Cori. Additionally, it is assumed that there is a bin directory at the same level as the wdl files which contains numerous scripts for different parts of the annotations. On Cori at NERSC, this directory is /global/dna/projectdirs/microbial/omics/gbp/img-annotation-pipeline/bin and can be copied here into this directory to be used in a relative way.

The workflow is written to take in a metagenome or isolate FASTA which is split into some number of smaller files. The directory structure looks like:

/path/to/splits/<split number>/file.fna

The splits should go from 1 to N, and N should be specified in inputs.json as the variable metagenome_annotation.num_splits. The workflow assumes that all FASTA splits have the same file name. So, for example, if we have an example FASTA example.fna split into 3, the directory structure would look like:

/path/splits/1/example.fna
/path/splits/2/example.fna
/path/splits/3/example.fna

where "example" should be set as the value of the variable annotation.imgap_project_id and /path/splits/ should be set as the value of the variable annotation.imgap_input_dir.

Instructions

Edit inputs.json to pick whichever parts of the annotation pipeline(s) should be run. Make sure that the locations of the input databases and executables are correctly set as well.

The annotation workflow is structured:

|annotation
|- setup
|- structural annotation
|-- pre-qc
|-- trnascan
|-- rfam
|-- crt
|-- prodigal
|-- genemark
|-- gff_merge
|-- fasta_merge
|-- gff_and_fasta_stats
|-- post-qc
|- functional annotation
|-- ko_ec
|-- smart
|-- cog
|-- tigrfam
|-- superfamily
|-- pfam
|-- cath_funfam
|-- signalp
|-- tmhmm
|-- product_name_assign

Run the workflow with the command:

java -jar <Cromwell> annotation.wdl -i inputs.json

in this directory.