Skip to main content

Table 1 HMMERHEAD Speed Improvement

From: Hidden Markov model speed heuristic and iterative HMM search procedure

Model

Algorithm

θ

δ

μ

η

Db Reduction

Fold Speedup

Local

Forward

6

2

7

20

6,805/2,313,578 = 0.3%

19,880s/957s = 20.8X

Local

Viterbi

6

2

7

20

6,805/2,313,578 = 0.3%

5,066s/858s = 5.9X

  1. Average search time for 500 benchmarking models was calculated using HMMERHEAD Forward or Viterbi with several different filtering thresholds. This average time was then compared to the average search time of default HMMER 2.5.1 Forward or Viterbi, 19,880 and 5,066 seconds, respectively. HMMERHEAD Forward provides the greatest speed improvement, ~24X, due the increased average time of this scoring algorithm.