neurospin / brainomics Goto Github PK

locate_genes work wrongly since that
for a given SNP
it scans all gene whatever the chromosome they (the genes) belong to.

This create erroneous links betwen this snp and several ones when only one is expected. The cardinality is wrong as well.

The entity Gene in the cube genomics has a relation chromosomes (plural !), with cardinality SubjectRelation('Chromosome', cardinality='+*').

This is not expected in real biological world.

I checked with this request :
"Any G,GN GROUPBY G,GN WHERE G name GN, G chromosomes C HAVING COUNT(C) > 0" -> got 23693 answer (OK )
"Any G,GN GROUPBY G,GN WHERE G name GN, G chromosomes C HAVING COUNT(C) > 1" -> got 0 (OK).

The DB is consistant with the real world.
This cardinality should be fixed.

Some genes are not assigned to a chromosome (hg19regfGene.txt.met). This should be fixed in import_ncbi1.sh (10 genes concernés...)