#1 COGreadblast symmetrize functionality not working?

[Sonia Timberlake]

I am using cogsoft to generate clusters in 93 genomes. My blast output is not fully symmetric, and I indicated this to COGreadblast with the "-r" option (also the -a option, because that was applicable). I had two nonoverlapping genome sets A and B. The blast output file was structured as following (querySet_vs_subjectSet): A_vs_A, A_vs_B, B_vs_B (concatenated from many blastp files generated with -m9). The resulting clusters included proteins only from set A OR from set B, but no cluster contained proteins from both sets. After appending blastp results from B_vs_A to the blast file and rerunning clusters, many clusters contain proteins from both sets, as expected from the phylogeny. I am guessing this means that COGreadblast symmetrize functionality is not working. I can see that A_vs_B hits are reported in hits.csv, but B_vs_A hits are not listed there, and by the time you get to all-edges.txt, the A_vs_B hits are gone as well.