MSOAR 2.0: Incorporating tandem duplications into ortholog assignment based on genome rearrangement

BMC Bioinformatics

Table 2 Comparison of the performance of five programs using gene symbol validation.

Pair of Species	Program	Assignable	Total Assigned	True Positives	Unknowns	Sensitivity	Specificity
human vs mouse	InParanoid	14,341	16,058	13,216	1,394	92.16%	90.13%
	Ensembl	14,341	20,670	13,619	2,850	94.97%	76.43%
	MultiZ	14,341	16,543	13,136	1,433	91.60%	86.94%
	MSOAR	14,341	16,769	13,528	1,554	94.33%	88.91%
	MSOAR 2.0	14,341	16,774	13,625	1,551	95.01%	89.50%
human vs mouse	InParanoid	12,688	15,197	11,750	1,529	92.61%	85.97%
	Ensembl	12,688	18,814	12,004	2,490	94.61%	73.54%
	MultiZ	12,688	16,102	11,600	1,570	91.42%	79.82%
	MSOAR	12,688	15,883	11,970	1,723	94.34%	84.53%
	MSOAR 2.0	12,688	15,942	12,085	1,765	95.25%	85.24%

In order to assess the accuracy of InParanoid, we take the first pair of genes in each ortholog group (i.e., the main ortholog pair of the group) as a one-to-one ortholog pair. For the Ensembl ortholog database, we directly download all the ortholog pairs from Ensembl Biomart Browser, which includes one-to-one, one-to-many, and many-to-many orthology relationships. In order to extract the orthology information from MultiZ, we download the whole-genome multiple alignment for human, mouse and rat from UCSC genome browser, and map the annotated genes to the alignment based on their coordinates on each genome.

ISSN: 1471-2105