[maker-devel] maker_coor behaviour

[Carson Holt]

I wouldn’t use those options for standard de novo annotation.  There are
really other more appropriate thing that should be used instead.  Both
maker_coor and est_forward are destined to be part of a separate tool that
will secretly just be calling MAKER, but will allow me to control what
other parameters MAKER sees to avoid certain logic incompatibilities that
make sense when mapping entire genes onto a new assembly, but not really
for de novo annotation using ESTs.

You should instead try modifying these options in the maker_bopts.ctl file
—>

pcov_blastn= #Blastn Percent Coverage Threhold EST-Genome Alignments
pid_blastn= #Blastn Percent Identity Threshold EST-Genome Aligments
eval_blastn= #Blastn eval cutoff
bit_blastn= #Blastn bit cutoff
depth_blastn= #Blastn depth cutoff (0 to disable cutoff). For trimming
high evidence overlap regions

en_score_limit= #Exonerate nucleotide percent of maximal score threshold


If either blastn or est2genome results disappear, it is because they don’t
meet one of these thresholds (blastn results that don’t meet the
thresholds but are borderline are kept if exonerate does meet the
thresholds, but if exonerate misses a threshold they will be thrown out).
That is whey the EST in question gets thrown out and it’s why the blastn
result disappears when you try and anchor it with maker_coor.

You can visualize everything with a browser when your done.  I still
recommend the old version of Apollo for this (it’s just easier).  You can
try and install it using the ‘./Build apollo’ option from the
.../maker/src/ directory, and it will be installed in
.../maker/exe/apollo.  It requires that you have apache ant installed to
do this.  Otherwise just download it from the GMOD source forge page and
install it manually.

Thanks,
Carson



On 2/28/14, 3:40 AM, "Mikael Brandström Durling" <mikael.durling at slu.se>
wrote:

>Hi,
>
>in a previous thread, the maker_coor feature for ETSs was mentioned. I
>have been trying it out, without using it for mapping gene names. I have
>placed these ESTs by other means, an thought the maker_coor feature would
>be a good use of this a priori knowledge. My major problem i try to solve
>is that I find that some ESTs where I know where they should be aligned,
>are not recruited to that position by maker’s blastn->exonerate method (I
>find them on other scaffolds). So I thought maker_coor with the
>est_forward behavior (as described) would be a good option to force my
>evidence onto the correct position, instead of ending up supporting or
>braking other models. However, as soon as I run with maker_coor tagged
>est sequences, no est2genome evidence appears in the final gff3 file. The
>blastn evidence is there when est_forward is disabled, but as expected,
>there is no blastn evidence when est_forward is turned on. It seems
>though as the evidence is used, as the QI lines indicate EST support for
>both splice sites as well as exon alignments, but I have no way to
>visualize and/or evaluate the congruence of evidence and models. Would it
>be possible to tweak Maker into outputting the est2genome alignments when
>est_forward/maker_coor is used? I couldn’t figure myself where in the
>code this was handled.
>
>I could of course do my own exonerate alignments of these ESTs and feed
>them into maker as est_gff, but if maker already has the machinery to to
>this, I thought it would be a good idea to use it.
>
>Thanks,
>Mikael
>
>
>_______________________________________________
>maker-devel mailing list
>maker-devel at box290.bluehost.com
>http://box290.bluehost.com/mailman/listinfo/maker-devel_yandell-lab.org