Skip to main content


Fig. 2 | Biology Direct

Fig. 2

From: xHMMER3x2: Utilizing HMMER3’s speed and HMMER2’s sensitivity and specificity in the glocal alignment mode for improved large-scale protein domain annotation

Fig. 2

No. of more sensitive HMMER2 models and sensitivity performance of HMMER2 and HMMER3 Pfam (release 27) model builds. a depicts the proportions of HMMER2 (red), HMMER3 (blue) and undetermined (green) model builds for the various sets of FP values. Based on the HMMER2 plot (in red), a stringent set of 3066 to 3502 HMMER2 models were found to be more sensitive than their HMMER3 counterparts at all evaluated FP values. This is 21.5% of the evaluated Pfam domain library (release 27). On the other hand, 68.5% (11158/16295) to 78.3% (12757/16295) of the HMMER3 domain models remain more sensitive than their HMMER2 counterparts (blue plot). Up to 12.7% of the domain library fluctuates between HMMER2 and HMMER3 depending on the evaluated FP value (green plot). b shows the average normalized AUC difference (in terms of percentage) for the sets of HMMER2, HMMER3 and undetermined models. The average AUC difference ranges between 1.61 and 3.59% for HMMER2 models (red plot) while it is held steadily at about 0.3% (blue plot) for HMMER3. For the sets of undetermined models, the average AUC difference stabilizes to about 0%

Back to article page