FeatureCollection::esource: invalid source key: E

[Admin]

Originally posted by njauxiongjie in the old forum on 18.03.2012 - 12:19

I have aligned my EST to my genome and want to use the EST alignment as hints of augustus. The command I run like your README file in AUGUSTUS, and like bellow:

Code: Select all

blat -minIdentity=92 genome.fa cdna.fa cdna.psl blat2hints.pl --in=cdna.psl --out=hints.E.gff

but when I feed the hint file to augustus, it appeared some error like bellow:

Code: Select all

augustus: ERROR
        FeatureCollection::esource: invalid source key: E

How can I solve this problem?

[Admin]

Originally posted by mario in the old forum on 18.03.2012 - 12:20

The problem is probably that you are using the default extrinsic.cfg file in the folder config/extrinsic that does not know the source key E.

Copy one of the extrinsic.cfg example files from this folder that contains E in the line after [SOURCES] to, say,
extrinsic.myspecies.cfg and then run

Code: Select all

augustus --extrinsicCfgFile=extrinsic.myspecies.cfg ...

[Admin]

Originally posted by Mbandi in the old forum on 06.10.2014 - 11:39

With external evidence available as hints, it appears AUGUSTUS will predict genes even in regions with not evidence. Is there a flag to turn off this behaviour?

[Admin]

Originally posted in the old forum by katharina on 07.10.2014 - 09:33

AUGUSTUS is build to predict genes ab initio (i.e. where there is no evidence at all), but if you add evidence, it will use the evidence where it is available. You can filter the predictions for genes without evidence (under each gene, you find a list: how many hints support a gene) and discard those if you don't want them.

[Admin]

Originally posted by Mbandi in the old forum on 09.10.2014 - 12:41

For genes that were predicted with hints, I would like to generate the list of all hints per gene. The output of AUGUSTUS seems to truncate the list of hints when it is long. For example
1. hint groups fully obeyed: 5
2. XNT: 5 (T_comp186006_c0_seq1_1,ol_SB13_comp208950_c0_seq1_1,_comp186006_c2_seq1_1,...)
Here I see only 3. Is there a way to specify that all the hints are written to the output?
What does "# % of transcript supported by hints" represent? is it the fraction of the hints group supporting the predicted gene?

[Admin]

Originally posted by mario in the old forum on 09.10.2014 - 17:10

We don't have a parameter for that (there are already >200 command line parameters) but you can simply edit the source code file src/gene.cc.
Go to the function
Evidence::print()
change the number in the expression
grouplistlen < 80
or remove this condition altogether from the while loop. Then recompile.
% of transcripts supported is
computed in the same source code file in the function
getPercentSupported
It is the percentage of exons and introns of the transcripts that have at least some evidence, the amount of evidence does not count.