Regions harboring copy number variants (CNV)
might also be under selection and contributing to an
observed selection signal, therefore CNV regions that
coincide to smoothed FST peaks were further explored.
Gene content of cattle CNV regions was assessed using
Ensembl (ftp://ftp.ensembl.org/pub/current_fasta/bos_
taurus/pep/). It is worthwhile to point out that FST and
CNV results did not use the exact same samples. CNV
results are based on Bickhart et al. [21] that use a
Holstein, a Nelore, a Hereford and 3 Angus samples,
also included in the FST analyses. Intersections between
balancing selection region coordinates and exon positions
were compared using MySQL queries. We obtained
a catalog of all bovine peptides from Ensembl.
This yielded 22,118 peptides, 345 of which overlap
with 24 predicted balancing selection regions, and
corresponded to 318 unique Ensembl genes. Using
PANTHER version 7 [22], we tested for over representation
of biological process, molecular function and
pathway terms under the balancing selection regions.
Results were Bonferroni [23] adjusted and PANTHER
terms with less than five observations were not further
analyzed. Similar analyses were performed on the
peptides under the 48 positive selection regions. PANTHER
results were similar when all peptides under the 24 balancing
selection regions and 48 positive selection regions
were combined in a single analysis.