A metaanalysis of bat phylogenetics and positive selection based on genomes and transcriptomes from 18 species

Hawkins, John A.; Kaczmarek, Maria E.; Müller, Marcel A.; Drosten, Christian; Press, William H.; Sawyer, Sara L.

both the exon-structure filtering step and the MIXR algorithm improve sequence conservation. Fig. 2F shows the positions and distribution of conserved sites in the first 2,000 multiple sequence alignments. Many of the poor quality exons are at the ends of the alignments, which is to be expected. The ortholog finding process scores sequences based on length of agreement, which naturally tends to include matching sections in the center. The ends are then more free to vary, with variability expected due to differences in isoform selection, as well as due to incomplete or incorrect transcript assembly. Transcripts tend to have less coverage near the ends and correspondingly poorer assembly. This also helps explain why so many unanimous sites end up being filtered: correctly matched exons will still be filtered if partial assembly results in exons of different lengths. From these results, as well as manual inspection, the alignments have significantly fewer erroneous columns after exon filtering.