Figure 6
From: Sequence similarity is more relevant than species specificity in probabilistic backtranslation
EasyBack Performace Analysis on FGA, TYR, MYH9 and HIST4H4. Left column: EasyBack prediction performance (percentage of amino acids correctly decoded). Input proteins are: FGA (fibrinopeptide), TYR (tyrosinase), MYH9 (myosin), HIST4H4 (histone H4). Right column: forward probability. The quality of prediction using BLAST All-Species training sets is higher than both Random Species-Specific (sequences belonging to the same organism) and BLAST Invertebrates (distant organisms). The forward probability can be used to estimate the best training set size. In almost all cases a high forward probability corresponds to a high quality backtranslation.