Closed hoelzer closed 4 years ago
basic blast and metadata-combine (ruby script) implemented for first tests with the nextflow pipeline.
CWL workflow in progress, blast - filter - merge .cwl steps reader. Ruby script migrated to python.
Missing the integration in the workflow.
https://github.com/EBI-Metagenomics/emg-virify-pipeline/pull/new/imgvr-workflow-step
@hoelzer ToDo: implement the python version of the merge script
@hoelzer there is one already in place https://github.com/EBI-Metagenomics/emg-virify-scripts/blob/imgvr-scripts/virify_scripts/imgvr_merge.py
Background
As additional validation step we can check how many of predicted viral sequences we can blast in a database of viral sequences including metagenomes: https://img.jgi.doe.gov/vr/
Idea
I just checked this again by blasting (megablast and blastn) vs. IMG/VR (including viruses from metagenomes so makes most sense to me) using 8 threads
query: 1590 contigs (all high confidence, low confidence and putative prophages from a virus enriched groundwater sample)
blastn (55 minutes) megablast (30sec)
Restrictive filtering of the blast results (aln length 80%) revealed
92/1590 (blastn) 33/1590 (megablast)
hits against IMG/VR that not really tell us some taxonomy but at least these are also reported (unclassified) viruses. The 33 are all included in the 92. For 5 of them, VIRify found some taxonomy.
So we could simply add this to the pipeline and provide some additional information about if this sequence was seen before as 'some virus'.
BLAST Database location currently: /hps/nobackup2/metagenomics/mhoelzer/nextflow-results/virify/v1/kallies_2019/IMG_VR/IMG_VR_2018-07-01_4
Downloaded from: https://genome.jgi.doe.gov/portal/IMG_VR/IMG_VR.home.html
commands
@mberacochea I would implement this in nextflow first and add the database to the ftp and I think a CWL translation is then straight-forward. The visualization of the hits is something to think of later