Annotation for contigs from de novo assembly

[witty]

Hi,

I want to annotate my assembled contigs (from de novo assembly). I used BLASTX and only got 10~20% percentage of hits(evalue=1e-5). Now all my differentially expressed contigs (genes) have no annotation. At least I want to know what these genes are, e.g, signaling, transmembrane etc.

Thanks a lot!
Victoria

[themerlin]

I'd give Prokka a try:

http://www.vicbioinformatics.com/software.prokka.shtml

[GenoMax]

Provided Victoria is working with a prokaryotic genome

NCBI has a eukaryotic annotation pipeline: http://www.ncbi.nlm.nih.gov/genome/a...n_euk/process/ and a prokaryotic one: https://www.ncbi.nlm.nih.gov/genome/annotation_prok/ If I recall right, you will have to make the sequence public though at some point in time if you use these.

Other eukaryotic options (have not used myself):

Pasa: http://pasa.sourceforge.net/
Maker: http://www.gmod.org/wiki/MAKER

[dnusol]

I think Blast2GO would also be useful

[ChrisAU]

I've also had good experience with Blast2GO, it doesn't require installation and is quite easy to handle. Also, they updated the quite ugly colours of their pie charts

[witty]

Hi,

Thank you for your reply. I understand that blast2go (see the below link) just used blast result so basically it won't provide more annotated contigs than BLASTX that I did, is it correct?

https://sites.google.com/a/brown.edu/bioinformatics-in-biomed/b2g4pipe-2-5

The organism I want to annotate is the protist, Oxyrrhis Marina.

Thank you!
Victoria

[Krish_143]

RAST annotation.

http://rast.nmpdr.org/

[dnusol]

Hi Victoria, I guess you could use several databases to increase your chances of annotation. What databases have you used? I don't have experience with protists but in general a good start could be to compare against GenBank and Uniprot's Swiss-Prot and TrEMBL protein databases. Have you tried a less conservative e-value? Also try to download similar species that are annotated to compare directly. This reference may help you

http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0014202

Dave

[Birdman]

You can try the Trinotate pipeline. It involves several tools (TransDecoder to get plausible ORFs, PFAM, HMMER, signalIP, tmHMM, RNAmmer) to obtain a quite complete annotation report. They give a lot of details on the website on how to use it.

[bastianwur]

Run a gene prediction tool (e.g. prodigal) over it, throw the proteins in InterproScan, and check if you get anything interesting for your analysis.

Might as well be good to know how long the contigs are.
Will not be of much use to annotate stuff, which is considerable less long than 900 bp.