Bug: Looks for the taxonomically closest organism for matching the Kcats
HossFir opened this issue · 8 comments
Description: There was an issue in fuzzyKcatMatching.m when it searches BRENDA for the closest relative organism when the Kcat is unavailable.
The type of error was:
The logical indices contain a true value outside of the array bounds.
I commented the line 493 to skip the issue.
@HossFir Could you please edit your post and provide more information in accordance to this template?
Would you be able to share the ecModel so that we can try to replicate the problem? This part of the code has been unchanged since GECKO2, so it should not suddenly have a bug.
Sure! This is the ecModel for my organism:
I cannot replicate the error. What did you specify as obj.params.org_name in the ModelAdapter? And what MATLAB version are you running?
obj.params.org_name = 'synechocystis sp. PCC 6803'
I am using MATLAB R2020b.
Using GECKO 3.0, I commented the below lines, and the ecModel has been reconstructed!
% if ~isempty(EC_indexes)
% distances = phylDist.distMat(org_index,KEGG_indexes);
% EC_indexes = EC_indexes(distances == min(distances));
% end
It is also worth mentioning that by using GECKO 2.0 pipeline, more enzymes were attached to the GEM.
Using GECKO 2, the number of proteins attached to the modes was: 589, while this for GECKO 3.0 is 444!
The difference in number of enzymes is probably because how the Uniprot data is gathered. How did you do this for GECKO 2, and what parameters do you set for GECKO 3 to do this automatically?
For GECKO 2, I followed the protocol given and downloaded the Prot_abundances from PaxDb and also retrieve Kcat values from BRENDA.
Using GECKO 3, I just set all these automatically in the ModelAdapter.m.
obj.params.org_name = 'synechocystis sp. PCC 6803'
The issue arose because there are multiple KEGG entries with the same species name (syn and syz). Is resolved in #289.
Regarding the number of proteins included, this is not related to PaxDB nor BRENDA, but really the data gather from Uniprot. This is indeed different between the GECKO versions, but you can open a separate Issue for this if desired.