Table 1 Feature set. Weighted means of all assessed features, and whether they were used in the final model.
From: Machine-learning approach expands the repertoire of anti-CRISPR protein families
Feature name | Acr mean | Non-Acr mean | Used in final model |
|---|---|---|---|
Containing Genome is Prokaryote | 0.8956 | 0.8958 | No |
Containing Genome is Self-Targeting | 0.3344 | 0.1919 | Yes |
Directon Annotated Protein Fraction | 0.22 | 0.69 | Yes |
Directon Protein Lengths Mean | 119.27 | 251.71 | Yes |
Directon Predicted Membrane-Associated Fraction | 0.06 | 0.26 | No |
Directon Size | 3.49 | 3.5 | Yes |
Protein is Annotated | 0.0641 | 0.6731 | Yes |
Protein has HTH-Downstream | 0.4008 | 0.1181 | Yes |
Protein is Predicted Membrane Associated (TMHMM, SignalP) | 0.0256 | 0.2781 | No |
Directon Spacing | 18.37 | 13.7 | No |
Protein Length | 104.11 | 245.54 | Yes |
Mean Hydrophobicity (Kyte and Doolittle) | −0.48 | −0.15 | Yes |
