Skip to main content


Fig. 8 | Biology Direct

Fig. 8

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. 8

Calibration procedure between HMMER2 and HMMER3 domain model builds. Stage 1: The “hmmsearch” results for the two HMMER variants are first produced on a positive set of α seed sequences of the model with the rest of the 16294 models as consensus sequences. Stage 2: A total of β glocal HMMER2 and γ local HMMER3 sequence-to-domain alignments are paired HMMER2/HMMER3 hits or orphaned accordingly. For the paired hits, the HMMER2 hit is reconstructed to mimic the local alignment of its HMMER3 counterpart to derived a new HMMER2 E-value. Stage 3: the orphaned and reconstructed E-values of the HMMER2 are partitioned into their positive and negative sets (see red histograms). The same is done for the HMMER3 (see blue histograms). For a particular E-value cutoff, the respective false-positive rates (FPR) and true-positive (TPR) rates can be quantified from the two sets of histograms

Back to article page