Skip to main content


Fig. 4 | Biology Direct

Fig. 4

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. 4

The xHMMER3x2 webserver workflow. The three-staged xHMMER3x2 workflow is shown. In the first stage of xHMMER3x2, N (3066 ~ 3502) HMMER2 and M (13229 ~ 12793) HMMER3 models can be utilized for the search (where N < <M) based on the sensitivity evaluation between HMMER2 and HMMER3 models in Fig. 2a. Upon the model selection, the respective hmmpfam (HMMER2) and hmmscan (HMMER3) runs are executed with the respective sub-libraries of HMMER2 and HMMER3 domain models. In the second stage, the two runs will result in K local-mode and L glocal-mode sequence-to-domain alignments respectively. The resulting K local-mode alignments will be searched against the same K domain models but in glocal-mode HMMER2 to perform the full domain re-alignment step. The third stage of xHMMER3x2 estimates the FPR(FP) values of the domain hits from their E-values using the previously calibrated HMMER2/HMMER3 ROCs. The FPR intervals are bounded by the FPR(FP) points: 6.14e-5 (1), 3.07e-4 (5), 6.14e-4 (10), 9.21e-4 (15), 1.20e-3 (20), 1.50e-3 (25), 1.80e-3 (30), 2.50e-3 (40), 3.10e-3 (50), 3.70e-3 (60), 4.30e-3 (70), 4.90e-3 (80), 5.50e-3 (90) and 6.14e-3 (100)

Back to article page