search

fmaguire / WoLFPSort

WoLFPSort v0.2 source code as original server no longer available
Other
9 stars 11 forks source link

Possible RCL1_YEAST misclassification in test data #2

Open billzt opened 8 years ago

billzt commented 8 years ago

I run the test data as: ./runWolfPsortSummary fungi <testQuery.fasta

The output is:

RCC1_YEAST nucl 12.5, mito_nucl 11.5, mito 9.5, cyto 3
RCL1_YEAST mito 15.5, cyto_mito 12.833, cyto 9, cyto_nucl 5.833
RT04_YEAST mito 16, nucl 8, cyto 3
EF1A_ASHGO cyto 27

It seems as if the result of RCL1_YEAST is wrong. In testQuery.fasta it shows that RCL1_YEAST is nucl, not mito. Other three are right.

fmaguire commented 7 years ago

I'm not 100% (not my project, just adopted it when the original host died and I couldn't find another copy of the source).

But in the REGRESSION file that Paul Horton (the developer) used for testing, the result you are getting are consistent with that:

RCL1_YEAST mito 15.5, cyto_mito 12.833, cyto 9, cyto_nucl 5.833

It's possible that this was deliberately chosen as a test example because it was a sequence that was classified incorrectly, but without knowing more about the origin of the sequence I can't really say.

Considering, the other test cases with the animal and plant networks are being run on the same fungal sequences who knows.

Maybe try e-mailing "Paul Horton paulh@iscb.org"?

paulhorton commented 7 years ago

I don't remember how I set up the WoLF PSORT package. It was quite a while back... However in general it is not unusual for WoLF PSORT to make some incorrect predictions, so that in itself is not a reason to worry that WoLF PSORT is not working normally.

PS. Big thanks to fmaguire for setting up this github repository! Recently I started to use github/gitlab myself and may be able to answer questions about WoLF PSORT.