Re: prioritizing est's?

classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view

Re: prioritizing est's?

Carson Hinton Holt
Re: prioritizing est's? Extending final models for starts/stops is one of the options added.  Unfortunately there is not a way to give some ESTs more confidence than others, but how EST’s and proteins are processed and compared to the gene models has changed somewhat, so some things will align better (after an improved polishing step) and there will be fewer spurious alignments that could lead to low confidence gene models.  There is also a GeneMark fix for a bug that was introduced in the last release.  A couple of other changes include an auto-install option, so MAKER will reach out over the internet and then install missing dependencies for you.  There is also support for comma separated lists, so you can now provide multiple files to each maker_opt value.  Along with the files you can also supply an optional label, so you can label some BLASTX alignments as UniProt, others as FlyBase, others as NCBI, etc.  This is a convenience feature for creating sub categories when using MAKER GFF3 with GBrowse, JBrowse, and Apollo.

My goal was to have this up today, but it looks like it will take a little longer.  I am still testing a few more thing, but I will put this up as a beta-version next to the current release probably this evening or tomorrow.


On 10/8/10 1:20 PM, "Kymberlie Hallsworth-Pepin" <kpepin@...> wrote:

Hi Carson,

Is there any way to prioritize a subseq of est's, say giving them more
confidence than the rest of the ests?

Also, any info on the good blast hits not being used in maker or if the
new release of maker you mentioned this week will give the option for

>I will try and reformat the hints to see if FGENESH behavior changes, I
>will also add an option to have MAKER walk out and find the start codon
>(a control file parameter), as well as an option to turn off hints to
>specific predictors so only the ab initio gene predictions are considered.
>I'll let you know when I'm done.


maker-devel mailing list
[hidden email]