Skip to main content
Fig. 3 | BMC Bioinformatics

Fig. 3

From: Combining genetic algorithm with machine learning strategies for designing potent antimicrobial peptides

Fig. 3

Widened parameter coverage for screened peptides through codon-representation. Generation 20 was shown to have higher scoring variation for codon-representation peptides in Fig. 4c. The spread of the peptides predicted to be antibacterial either generated by the codon representation method (teal) or the non-codon representation method (red) is shown. Each contour is a change in density of 0.01. The 2-D density contours show that the codon-representation generates screened peptide sequences which have AGGRESCAN scores outside the non-codon representation at most peptide lengths, both above and below, even though the non-codon representation generated more screened sequences as seen in Fig. 4a. The codon-representation also generated the majority of peptides at the extremes of peptide length, 6 and 13

Back to article page