HELP: GO analysis of non-natively supported organism using R package goseq

[tianyub836]

Hi,all
i am trying to perform GO analysis with goseq, but the organism in my work is not natively supported by the package and i am either not able to get myself understood the steps in the package manual.

so i post here to see if anyone who happens to know the analysis procedures for non-natively supported organism using goseq. any helps about other softwares or packages for go analysis are appreciated.

thanks

[poisson200]

Hi.
You may want to try the Bioconductor mailing list if you have not already and ask there.

https://stat.ethz.ch/mailman/listinfo/bioconductor

If you don't find an answer. You could find all orthologs between your genes and a model species (one that is supported by GOseq). Then for each ortholog found, categorised as being differentially expressed or not based on your data and then use GOseq.

A simple reciprocal best hit approach using BLAST will find orthologs.

Of course, others may have better/convenient solutions.

Good luck,

Cheers,

John.

[tianyub836]

Thanks for your reply, John.
Actually, i have obtain the GO identifiers corresponding to my genes/transcripts. The issue i have met is that i do not completely understand the basic principles of GO analysis using goseq and as a result of that i am not able to generate the data structures needed for the package.

As to the orthologs, actually i have checked that and found that it seemed that there is no orthologs of my organism in the supported list of goseq.

The alternative i am considering about is writing my own scripts base on the principles of GO analysis if there is no tools suit for my work, but i am a newer of GO analysis and so it may take much time.

[AliceRNA]

bad nullp plots for pwf

Hi John,
I was using GOseq for a non-model species, inspired by your comments. I was using length generated from scratch and also I want to import the best hits of Arabidopsis. I was using nullp to generate a pwf however, I found that the fitting curve is bad, as there are only 8 points showing on the nullp plots. However, in the Goseq manual, the curve fit well with many points. Also the manual showed that there are about 300 gene bins, yet my plot has 6100 gene bins.

Is there any idea in terms of why this happens? Also any thoughts on how to change parameters to fix this?

Thanks.

Looking forward to the reply.

AliceRNA