[maker-devel] Alternative splicing in MAKER

[Carson Holt]

est2genome=1 together with alt_splice=1 can cause weird behavior, because est2genome is just a cut and paste of an alignemnt to being a gene model, it will always be 100% supported by the evidence (itself as an alignment), and anything that overlaps will be clustered to being the same gene which can be messy if models you are moving forward align to multiple locations. 

You can add est_forward=1 (manually add it, it’s undocumented) to maker_opts.ctl to get MAKER to do a few extra behaviors. It will keep the names from the est2genome alignments (not rename them to maker names), and if you add hints like gene_id=<gene_name> to the fasta header it will only cluster things with the same gene ID and not just cluster by overlap. Also you can add maker_coor= to the header to restrict alignments to specific contigs or even contig regions.

—Carson



> On Feb 9, 2020, at 3:24 AM, Lior Glick <liorglic at mail.tau.ac.il> wrote:
> 
> Hello,
> I am working on a computational pipeline which involves genome annotation. Based on helpful advice I got in this mailing list before, I make two consecutive runs: the first is a liftover run with est2genome=1 and no ab-initio prediction, while the second run takes liftover results and adds ab-initio predictions, supported by protein and transcript evidence.
> In both runs, I get results which I find confusing regarding alternative splice variants prediction, but the behavior is different in each run.
> 
> In the liftover run, I use est2genome=1, alt_splice=1 and no ab-initio preduction.
> The resulting gff indicates many overlapping genes, coming from ESTs (transcripts actually) of different splice products of the same gene. Of course MAKER has no way to know that, but I was expecting that since the genes are highly overlapping, they will be grouped together as different mRNA features under the same gene.
> In the second run, I use est2genome=0,  alt_splice=1 and Augustus for gene prediction. Results of the liftover run are provided to the pred_gff parameter. In this case, it seems that overlapping genes are squished together, so I only get one gene with one mRNA.
> Please find attached maker_opts.ctl files for both runs, and GFF files demonstrating the issue (one gene example).
> 
> Could anyone please explain how this works? Why is the behavior different between the runs? Any way to get MAKER to behave the way I expected?
> 
> Thanks a lot!
> Lior
> <annotation.gff><annotation_maker_opts.ctl><liftover.gff><liftover_maker_opts.ctl>_______________________________________________
> maker-devel mailing list
> maker-devel at yandell-lab.org
> http://yandell-lab.org/mailman/listinfo/maker-devel_yandell-lab.org

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://yandell-lab.org/pipermail/maker-devel_yandell-lab.org/attachments/20200226/ab72f446/attachment-0004.html>