Gene fragmentation in bacterial draft genomes: extent, consequences and mitigation

BMC Genomics

Table 1 Effect of genome quality on annotation efficacy

Uncorrected
		Pfam^a	COG^a	KEGG^a
Including PP-C42	% Partial ORFs fragments vs. % all ORFs annotated	r = -0.854 P < 0.001	r = -0.551 P = 0.012	r = 0.586 P = 0.007
	Mean ORF length vs. % all ORFs annotated	r = 0.785 P < 0.001	r = 0.526 P = 0.017	r = -0.403 P = 0.078
Excluding PP-C42	% Partial ORF fragments vs. % all ORFs annotated	r = -0.421 P = 0.073	r = -0.019 P = 0.939	r = 0.415 P = 0.078
	Mean ORF length vs. % all ORFs annotated	r = 0.406 P = 0.084	r = 0.157 P = 0.520	r = -0.016 P = 0.949
Corrected using matched partial ORF sets
		Pfam	COG	KEGG
Including PP-C42	% Partial ORFs fragments vs. % all ORFs annotated	r = -0.861 P < 0.001	r = -0.595 P = 0.007	r = 0.469 P = 0.050
	Mean ORF length vs. % all ORFs annotated	r = 0.787 P < 0.001	r = 0.563 P = 0.012	r = -0.284 P = 0.253
Excluding PP-C42	% Partial ORF fragments vs. % all ORFs annotated	r = -0.338 P = 0.170	r = 0.027 P = 0.915	r = 0.350 P = 0.168
	Mean ORF length vs. % all ORFs annotated	r = 0.378 P = 0.122	r = 0.155 P = 0.538	r = 0.052 P = 0.842

^aPearson correlations between annotation frequency and genome quality, as represented by the percent of the predicted ORFs composed of partial sequences and mean ORF length. Complete genomes are excluded in all cases; including them has essentially no effect.

ISSN: 1471-2164