Table 2.
Antimicrobial prediction for the R23R, R23L, R23L*, and R23L* peptides.
| Peptide | CAMPR3 | AmpGram | AMP Scanner | |||
|---|---|---|---|---|---|---|
| RF | SVM | ANN | DA | RF and n-Grams | DNN | |
| R23R vs. R23R* | ||||||
| RKKRRQRRRGGGGLHITDMAWKR | 0.48 (non-AMP §) | 0.03 (non-AMP) | AMP §§ | 0.93 (AMP) | 0.59 (AMP) | 0.07 (non-AMP) |
| RKKRRQRRRGG-Sar §§§(A)-GLHITD-Nle §§§§(M)-AWKR | 0.49 (non-AMP) | 0.03 (non-AMP) | AMP | 0.95 (AMP) | 0.54 (AMP) | 0.05 (non-AMP) |
| RKKRRQRRRGG-Sar(A)-GLHITD-Nle(L)-AWKR | 0.58 (AMP) | 0.05 (non-AMP) | AMP | 0.96 (AMP) | 0.63 (AMP) | 0.03 (non-AMP) |
| RKKRRQRRRGG-Sar(P)-GLHITD-Nle(M)-AWKR | 0.48 (non-AMP) | 0.04 (non-AMP) | non-AMP | 0.95 (AMP) | 0.47 (non-AMP) | 0.13 (non-AMP) |
| RKKRRQRRRGG-Sar(P)-GLHITD-Nle(L)-AWKR | 0.53 (AMP) | 0.07 (non-AMP) | AMP | 0.97 (AMP) | 0.62 (AMP) | 0.07 (non-AMP) |
| R23L vs. R23L* | ||||||
| RKKRRQRRRGGGGITDFGIFIGL | 0.59 (AMP) | 0.06 (non-AMP) | AMP | 1.00 (AMP) | 0.37 (non-AMP) | 0.93 (AMP) |
| RKKRRQRRRGG-Sar(A)-GITDFGIFIGL | 0.59 (AMP) | 0.06 (non-AMP) | AMP | 1.00 (AMP) | 0.22 (non-AMP) | 0.44 (non-AMP) |
| RKKRRQRRRGG-Sar(P)-GITDFGIFIGL | 0.58 (AMP) | 0.10 (non-AMP) | AMP | 1.00 (AMP) | 0.45 (non-AMP) | 0.95 (AMP) |
§ It is predicted as a peptide that does not exhibit antimicrobial activity. The prediction level is less than 0.5. §§ It is predicted as a peptide with antimicrobial activity. The prediction level is over than 0.5. §§§ For calculations, sarcosine (Sar) was replaced by analogs similar in properties and structure: alanine (A) and proline (P). §§§§ For calculations, norlecine (Nle) was replaced by analogs of similar properties and structure: methionine (M) and leucine (L). Antimicrobial activity was predicted using algorithms: RF is a random forest, SVM is a support vector machine, ANN is an artificial neural network, DA is a discriminant analysis, RF and n-gramms is a random forest based on amino acid motifs, DNN is deep neural network.