[maker-devel] maker gene prediction and overlapping genes

[Michael Thon]

Hi all - 

We've been having the same problem. In every case I've examined manually the overlapping gene models have overlapping CDSs and they are on opposite strands. In most cases its easy to see which is the correct model because one has protein or EST/RNA-Seq evidence and the other does not. Most times one model is from Augustus and the other is from genemark, but not always. I found one in which both gene models were from augustus and maker promoted both of them.

I count 121 overlaps in our annotation (its a fungal genome). We're about to just go in and remove them manually but I want to see if there is any way to fix my configuration of maker first.

Mike



> On Sep 24, 2015, at 8:02 PM, Daniel Ence <dence at genetics.utah.edu> wrote:
> 
> Hi Anjuli, 
> 
> The approach that you outlined sounds pretty reasonable, and I’m not certain I understand the problem with your results. Are the short genes that lie completely in other genes in the introns? Or do you mean that you have overlapping predictions? 
> 
> A common observation in compact fungal genomes is that maker can produce gene models that fuse several adjacent genes together. Could that be what you’re observing? There's actually an option in maker to deal with that issue; it’s the “correct_est_fusion” setting in the opts control file. 
> 
> Let me know whether that helps, 
> Daniel
> 
> Daniel Ence
> Graduate Student
> Eccles Institute of Human Genetics
> University of Utah
> 15 North 2030 East, Room 2100
> Salt Lake City, UT 84112-5330
> 
>> On Sep 23, 2015, at 2:08 AM, Anjuli Meiser <anjuli.meiser at gmail.com> wrote:
>> 
>> Hello,
>> 
>> I am using Maker in two rounds for gene prediction in fungal genomes.
>> 
>> In the first round I'm running maker with the HMMs gained from GeneMark and snap with hints from CEGMA and include RNA evidence through a tophat gff. Then I convert the maker results to new snap HMMs and augustus HMMs and run maker in a second round. I also rescue rejected gene models (maker standard build) by running interproscan.
>> 
>> I observed that I get around 10-15% of genes that are overlapping in some way. That includes short genes predicted to lie completely within the boundaries of larger genes and also normally overlapping (mostly on opposite strands).
>> 
>> Do you have a suggesting how to deal with this? Did I miss some settings in maker to reduce these or at least filter out the shorter genes that are lying within other genes?
>> 
>> Thank you very much in advance for any help in this matter!
>> 
>> Best wishes,
>> Anjuli
>> 
>> _______________________________________________
>> maker-devel mailing list
>> maker-devel at box290.bluehost.com
>> http://box290.bluehost.com/mailman/listinfo/maker-devel_yandell-lab.org
> 
> _______________________________________________
> maker-devel mailing list
> maker-devel at box290.bluehost.com
> http://box290.bluehost.com/mailman/listinfo/maker-devel_yandell-lab.org