17 months ago by
University College London
OMA groups (http://omabrowser.org/oma/landOMA/) contain sets of genes in which each pair are orthologous to one another. Consequently these groups contain at most one sequence per species. However, these sequence are not necessarily single-copy genes and thus the relationships are not all necessarily 1:1. But depending on what you are trying to achieve (e.g. marker genes for phylogenetic tree inference), these groups may be just as good or even better than sets of 1:1 orthologs.
If you really need 1:1 ortholog groups, we don't provide this specific output, but I can think of two way these could be reconstructed with a bit of post-processing:
1) retrieve pairwise orthology relationship files, remove all non 1:1 relationships, and then build groups using the transitivity of 1:1 orthology.
2) retrieve all hierarchical orthologous groups (HOGs) at the level of the mammals, and verify that there are at most one sequence per species. The HOGs that pass this filter are very likely to be 1:1 orthologs. For this, you can download all HOGs (http://omabrowser.org/All/oma-hogs.orthoXML.gz) and parse the file using the "Family Analyzer" tool currently under development in our lab: https://github.com/DessimozLab/familyanalyzer