SequenceBouncer: A method to remove outlier entries from a multiple sequence alignment

Abstract

Phylogenetic analyses can take advantage of multiple sequence alignments as input. These alignments typically consist of homologous nucleic acid or protein sequences, and the inclusion of outlier or aberrant sequences can compromise downstream analyses. Here, I describe a program, SequenceBouncer, that uses the Shannon entropy values of alignment columns to identify and remove outlier entries in a manner responsive to overall alignment context. I demonstrate the utility of this software using alignments of mammalian reference mitochondrial genomes, bird cytochrome c oxidase-derived sequence barcodes, and COVID-19 sequences.