Genome Annotation Pipeline Help Required

[marct]

First post as a user here, so please go easy on me for lack of due diligence

We have a genome assembled from Illumina data. There is a reference genome of a closely related species (same genus). I downloaded the proteins from this reference genome and sought to map them to our genome using local tblastn, as a homology-based annotation (we also have predicted transcripts from MAKER as an ab initio annotation method).

I have seen this method used in the literature, but all the method descriptions skip an important step - actually physically mapping the best tblastn hits (from whatever criteria) to the genome.

I assume there is some way to convert the blast xml output to an annotation file (GFF or similar) - one that conserves the info from the blast (especially protein name and function). I tried looking into Biopython and BioPerl but could not lay hands on the proper method of doing this.

Can someone please point me in the right direction?

[whataBamBam]

First post as a user here, so please go easy on me for lack of due diligence

We have a genome assembled from Illumina data. There is a reference genome of a closely related species (same genus). I downloaded the proteins from this reference genome and sought to map them to our genome using local tblastn, as a homology-based annotation (we also have predicted transcripts from MAKER as an ab initio annotation method).

I have seen this method used in the literature, but all the method descriptions skip an important step - actually physically mapping the best tblastn hits (from whatever criteria) to the genome.

I assume there is some way to convert the blast xml output to an annotation file (GFF or similar) - one that conserves the info from the blast (especially protein name and function). I tried looking into Biopython and BioPerl but could not lay hands on the proper method of doing this.

Can someone please point me in the right direction?

I'm currently doing something similar using gmap.. I have a transcriptmome that I'm mapping to a genome though.. (we assembled the transcriptome and then the genome of a related species was subsequently released) also have ests that I'm mapping to a genome with gmap. Exonerate I believe does a similar job and has a protein matching mode.. Anyway both these programs will output in gff format

[marct]

Thanks for the reply. As of this moment I am running genBlast and exonerate as well as the tblastn. In all of these cases, I am using the protein database of the model species as the query and my genomic sequence as the target (or database). I'll let you know how it goes.

[AdrianP]

How come the functions of MAKER were not sufficient for your analysis?

[marct]

I am joining this project somewhat in the middle of the process.

From my understanding, the initial Maker run was ab initio only, we do not have ESTs or RNA-seq data to add to the pipeline. So while important, the SNAP/Augustus etc gene calls from Maker should constitute one line of evidence for our annotations, while direct alignment of homologous proteins coupled with splice-site detection (a la exonerate) should constitute another, homology-based line of evidence.

Stop me if I'm wrong.

[whataBamBam]

Hang on..

I thought MAKER was just for ab initio. You can use that to bring together ESTs and RNA-Seq data too?

That's what I have. I'm just mapping my transcripts to the ESTs at the moment