View Single Post
Old 04-19-2013, 08:28 AM   #6
rhinoceros
Senior Member
 
Location: sub-surface moon base

Join Date: Apr 2013
Posts: 372
Default

Quote:
Originally Posted by Gorbenzer View Post
Local would be great, we have two servers for our analysis.

I'd like to try all the different approaces:
- Try to assembly all the data to see if i can get draft genomes of a few highly present species;
- Try to assembly all the data, predict ORFs and annotate them to obtain the gene pool of the sample;
- Try to "annotate" all the single reads to find out what genes are higly represented in the sample.

I would love to try gene prediction and annotation (including GO, EC and pathway prediction if possible) softwares specific for metagenomic analysis.

Thx for the help!
Well in this case, you're pretty much doing what I'm doing then. I use FragGeneScan for predicting protein-coding genes from contigs. These I've blasted against ncbi's nr database. For tRNA prediction I used tRNAscan-SE (3 runs each optimized for one domain). rRNAs by blastn megablast against Silva's SSU and LSURef databases. The M5nr database will be updated quite soon. If you want to get GOs, KEGGs and such with just one search, I recommend you find out more about it. If you do your blasts with -outfmt 5 or 6 (recommended) flags, you can import your data into MEGAN for neat analyses and visualizations. For other ideas, perhaps look into what e.g. IMG/M and MG-RAST do in their pipelines.. don't forget to screen for human contamination..

Last edited by rhinoceros; 04-19-2013 at 08:38 AM.
rhinoceros is offline   Reply With Quote