Re-Annotate legacy gene predictions

[Amative]

Hello everyone,

I have a number of single-ended samples generated by Illumina-based Hsq2000. About 10 to 35 million 58-bp. A legacy annotation of an organism "genemark predictions". What are my option to redo the annotations using the short read files I have.

Any tools, steps , suggestions, examples are really appreciated

Thank you!

[jimmybee]

Either de-novo assemble all SE HiSeq reads and predict genes on the assembled contigs, or map those reads against the current annotation and identify where improvements can be made

Denovo: Velvet, ALLPATHS, SOAPdenovo
Mapping: BFAST, Bowtie(2), BWA
Gene Prediction: FGENESH, GenMark, Genscan, Glimmer ....

[Amative]

Thanks Jimmy,
I am not quite sure I understand exactly how to do your second suggestion. Because I have already done the alignment to the genome and that is what caused my question.

Here is how this whole re-annotation idea came up:
I aligned the SE reads (different time points of fruit development and different tissues) to the reference genome. Then viewed the (Genome + Old predicted genes + Alignment results "BAM") using GBrowse. At this point we noticed that the reads are not always aligning perfectly to a number of the genes.

[Wallysb01]

I would highly suggest a two pronged strategy for use inside maker.

1) Use tophat and cufflinks RABT annotations to do transcriptome assembly on the genome.
2) Use trinity to de novo assemble the reads into transcripts.

Then reannotate your genome inside maker. You will be able to pass the legacy annotation through, along with refseq alignments from other species or a variety of other lines of evidence along with your de novo and reference based transcriptome assembly.

Finally, update your maker annotations with PASA using your de novo assembled transcripts.

[Amative]

Thanks a lot Wally
Sounds interesting, and a lot of work. I have never used any of the tools you suggested and excited to do so. Do you know of any links or documents and that list these steps with more details (not the manual of each), as I am no expert and need as much data as possible about this pipeline.

Have a good weekend

[Wallysb01]

Thanks a lot Wally
Sounds interesting, and a lot of work. I have never used any of the tools you suggested and excited to do so. Do you know of any links or documents and that list these steps with more details (not the manual of each), as I am no expert and need as much data as possible about this pipeline.

Have a good weekend

Yeah, I did something similar for this paper: http://www.biomedcentral.com/1471-2164/14/49

I did not feed legacy annotations to Maker though. Instead I merged Ensembl and NCBI annotations in EVM then fed the merged annotations to Maker.

If I were to do it again though, I'd probably have just skipped EVM and fed both Ensembl and NCBI into Maker.

[Amative]

Excellent, Thanks Wally.
I will definitely take a look at it, and hopefully I can do something similar.