multiple alleles: bug in mapAllelesCFtable?

[cecileane]

@crsl4 : would you have time to look into this? I don't unfortunately...

See google group question and attached file, thanks to @brunoasm : it seems that mapAllelesCFtable! would return an empty list of repeated species, such that mergeRows() is not called during readTableCF! and the resulting DataCF object still has 13,757,814 4-taxon sets instead of a few 1000s:

CF = PhyloNetworks.mapAllelesCFtable("pnetworks_pops.txt","all_cf.csv") # 13,757,814 rows × 14 columns data frame
dataCF_summarized = PhyloNetworks.readTableCF!(CF) # DataCF, number of quartets: 13757814

yet mergeRows worked:

dataCF_summarized = PhyloNetworks.readTableCF!(PhyloNetworks.mergeRows(CF, [1,2,3,4,5,8,11,14]))# 2375 unique 4-taxon sets were found.

(used PhyloNetworks v0.11.0 and julia v1.5.1)

[crsl4]

Sorry, I cannot find the attached file. Can you forward it to me to be able to check?
I am spread too thin, but I'll try to carve out some time.
Thanks!

[cecileane]

The file is attached to the google group question. It is not replicable (the data set is huge: 13,757,814 four-taxon sets), but I copy-pasted the key lines above. The problem is that dataCF_summarized still has 13,757,814 rows, instead of a few 1000s.

[brunoasm]

Hi @cecileane and @crsl4, thanks for looking into this!

Here I am attaching a subset of the data to reproduce the problem and help solving it.
pnetworks_test.zip

I also added a row in which all alleles map to the same species. readTableCF!() removes that one, but does not average rows with same species.

Explicitly calling PhyloNetworks.mergeRows() does average over rows.