We believe that this is fast enough for interactive use (instead of pre-computing BLAST hits for every query).
Synteny mapping approaches are too numerous to list, however most involve computing reciprocal best BLAST hits on putative ORFs, with BLAST hits filtered by e-value thresholds, coverage thresholds, and uniqueness criteria.
FastBLAST computes a BLAST-like alignment score for the pairwise alignments between the query and its homologs that are implied by the families' alignments, and it selects the top 2.5h homologs, where h is the desired number of top homologs and 2.5 is an arbitrary safety factor.
Therefore, for all supergenes that are not declared orthologs at a particular node of the evolutionary tree, we compute PHOG-BLAST scores to PHOG supergenes and for each such supergene we find the best hit.
Different tree views were computed based on BLAST pairwise alignment of the query sequences to the sequences searched in the databases (see Fig. 1 as an example).
Identity, contiguous matches, and near-contiguous matches between a probe and a target/non-target genome are computed from the BLAST alignments between the two.
We computed all pairwise BLAST comparisons (BLASTp, tBLASTx, BLASTx, and tBLASTn, depending on the type of database) at a threshold of e-value < 0.001 and counted the number of sequences with hits in the other species.
Last, like BLAST, computes local alignments based on a seed-and-extend technique, but it is fast because of the use of adaptive seeds thus allowing the application of classic sequence alignment techniques to the problem of mapping giga-scale sized sets of reads generated by high-throughput sequencers.
It took about 10 hours to compute the consolidation graph (including time consuming blast computations) and only several minutes to perform the refinement procedure.
We used BLAST to compute a reference of truth for the metagenome benchmark.
