OMA: gene_function.gaf output file empty

Question

[lassancejm]

Hi,

I am testing OMA version 2.1.1 on a Linux cluster. The program works fine with both the toyexample dataset and my own data. However, setting DoGroupFunctionPrediction := true; in parameter.drw produces an empty gene_function.gaf file (with the exception of the header).

No error message in the std out file, only warnings that taxa cannot be mapped to a clade (below is the corresponding output obtain with the ToyExample:

ReadLibrary: /n/home01/lassance/Software/OMA.2.1.1/OMA/OMA.2.1.1/lib/FunctionPropagation
{--> enter module, args = 
<-- exit module = proc (new_terms, fn:string) local i, genome, tax_id, j, e, id,
a; OpenWriting(fn); printf(!gaf-version: 2.1
); printf(! Functional prediction from OMA standalone using OMA groups.
); for i to NG do DB := DBs[i]; genome := genomes[i]; tax_id := GS[genome,
TAXONID]; if tax_id =  then tax_id := string(-1) fi; for j to GS[genome,
TotEntries] do e := Entry(j); id := sprintf(%s:%05d,genome,j); for a in 
combine_terms(SearchTag(GO,e),new_terms[i,j]) do write_gaf_entry(tax_id,e,id,op(
a)) od od od; OpenWriting(previous) end}
{--> enter PropagateFunctionByGroup, args = [[0, 0, 15, 13], [0, 0, 13, 11], [0,
0, 16, 14], [0, 1, 7, 6], [0, 6, 6, 5], [10, 0, 21, 19], [0, 5, 4, 3], [0, 0, 8,
7], [9, 17, 0, 0], [0, 14, 5, 4], [0, 0, 1, 1], [0, 0, 17, 15], [8, 18, 20, 18],
[0, 0, 9, 8], [0, 8, 11, 9], [0, 0, 12, 10], [0, 2, 2, 2], [0, 0, 18, 16], [0, 0
, 14, 12], [0, 4, 19, 17]], Tree(Leaf(dog,-41.833333,1),0,Tree(Tree(Leaf(mouse2,
-41.282218,4),-41.051115,Leaf(mouse,-41.051115,3),0.33429712),-40.630449,Leaf(
human,-43.166667,2),1))
WARNING: Unable to map dog to a clade. Will not propage function.
WARNING: Unable to map human to a clade. Will not propage function.
WARNING: Unable to map mouse to a clade. Will not propage function.
WARNING: Unable to map mouse2 to a clade. Will not propage function.
dog does not belong to a valid clade.
human does not belong to a valid clade.
mouse does not belong to a valid clade.
mouse2 does not belong to a valid clade.
Propagating function through group 20/20
  Predicted 0 annotations across 20 groups.
  Prediction took 0.00 seconds.
<-- exit PropagateFunctionByGroup = [table([],{}), table([],{}), table([],{}), 
table([],{})]}

If the option is only available with specific datasets it would be good to know. Or am I missing something to get the function prediction to work properly?

Advice/suggestion most welcome!

JML

Answer 1

[Adrian Altenhoff]

Hi,

indeed, the documentation is missing an important information. Function prediction in OMA is done on transferring existing annotations within OMA groups. The annotations can currently only come from exported genomes (from the omabrowser https://omabrowser.org/export ). We will add this information more explicit to the documentation.