Skip to main content

Table 4 Mean and median percent error for all methods on the benchmark without query RNA flanks (3,429 datasets).

From: Searching for evolutionary distant RNA homologs within genomic sequences using partition function posterior probabilities

Mean and median error Probalign SSEARCH BLAST ClustalW HMMER
Complete benchmark 30.8 | 30.4 31.4 | 22.1 32.0 | 20.9 37.9 | 38.5 44.9 | 44.7
Datasets with pairwise sequence identity at most 30% (14) 62.4 | 59.5 70.8 | 94.5 78.4 | 100.0 74.5 | 97.5 96.7 | 100.0
  1. Probalign has lowest mean and median error on low sequence identity datasets.