trainGlimmerHMM Error while training a genome

Question

[swati.6783]

I am using GlimmerHMM to identify the protein coding genes in a fungal genome. As the present trained set do not contain any fungal genome so I have to train my own model. In order to do this I am running the trainGlimmerHMM program.

It is giving me following error which I am unable to troubleshoot,

sh: line 1: 725252 Segmentation fault (core dumped) /HOME/swatis/bin/GlimmerHMM/train/build-icm < orfs_0_100 > coding_0_100.model ERROR 74: /HOME_ann/swatis/bin/GlimmerHMM/train/build-icm exited funny: 35584 at ../../train/trainGlimmerHMM line 614.

Any help!!

Answer 1

[wn1364]

i got the same error with u. then i checked my inputfile ,found the my gff file format was wrong。

Answer 2

[congliu37]

Hello,

I got the same error using ~20,000 genes for training and it was solved by reducing training gene set. I assume GlimmerHMM was not designed to take large training set, but reducing training set from 5,000 genes to 1,000 genes did lead to a drop in performance.