|
Overview:
Alignment algorithms such as BLAST typically return multiple hits
between
a given sequence and a collection of sequences. Correctly assigning the
best association between them, while avoiding spurious matches (often
pseudogenes or paralogs) that are "attractive" to standard alignment
tools, is an important starting point for making accurate biological
inferences.
With the move towards representing genomes and transcriptomes with only
draft-quality coverage, resarchers must wrestle with fragmented and
incomplete representations of the underlying data, a scenario which most
existing alignments techniques are not designed to handle.
To overcome these problems, EXOR applies multi-stage filtering to
exclude
lower scoring alignments and pick the highest-scoring mutually exclusive
alignments. This approach defines position-specific reciprocal best
alignments between two sequence sets, eliminating redundant and
competing
alignments and simplifying further analyses.
|
