Fig. 5
From: Searching for universal model of amyloid signaling motifs using probabilistic context-free grammars
Performance comparison of generalized BASS grammars and profile hidden Markov models on bacterial and fungal test motifs from NLR proteins. Performance in terms of recall at the false positive rates of \(1e{-}3\) obtained using the score averaging scheme. For fungal ASMs, the recall is averaged over the three classes of motifs (sigma, PP, HET-s). Data for PCFGs learned without (with) contact constraints are plotted in blue (magenta). Number of variables (l: lexical, s: structural) in grammars is indicated. Data for pHMMs learned with rough (accurate) MSAs are plotted in golden (chocolate). Method for generating MSA is indicated ([egl]insi refer to Mafft modes). See main text for details