How to map predicted proteins to contigs?

Question

[Paula]

Hi All!

I am looking for functional genes in a metagenomic dataset. I took the contigs assembled with metaSPADES and did protein prediction using Prodigal. Then, I used HMMER to search for the predicted proteins for the gene mcrA. Now I want to do tree placement, and the tree is based on DNA sequences. How can I search for the proteins annotated as mcrA in my contigs?

Thank you!

Answer 1

[Mensur Dlakic]

Let's say your metagenome assembly is in meta.fasta:

prodigal -i meta.fasta -a meta.faa -d meta.fna -o /dev/null -p meta

That command predicts proteins in meta.faa and matching genes in meta.fna, which are linked by a common header line. For proteins it may look like this:

>k141_164538_3 # 343 # 567 # 1 # ID=263_3;partial=00;start_type=ATG;rbs_motif=GGA/GAG/AGG;rbs_spacer=5-10bp;gc_cont=0.364
MGRTKKVGTAGRFGSKYGKKIREKVAEIEKIEKQRHICPNCKMRYLVREGTGIWVCKKCG
AKFAGQAYYPPRVS

That particular protein was translated from the third predicted gene in contig k141_164538 (thus k141_164538_3), it is between bases 343-567, in first reading frame. It is complete (that's what partial=00 means), starts with ATG, has a regular ribosome binding motif in front of it, and a G/C content of 0.364. If you identified that protein as your match, you just go to meta.fna and search for k141_164538_3 and that will be the gene you want.

[Paula]

Thank you, Mensur! This is incredibly helpful. Merry Christmas and Happy New Year!