Table 1.
Consistent | Ambiguous | <sd> b (φ/ψ) | Rmsd c (φ/ψ) | ||
---|---|---|---|---|---|
| |||||
Total | Bad | Warning | |||
For chemical shift database | |||||
| |||||
TALOS+ | 84.5% d | 4.1% e | 15.5% d | 12.4/11.9 | 13.6/12.5 |
TALOS-N | 90.7% d [86.6%/4.1%] | 3.5% e [2.9%/21%] | 9.3% d | 8.7/8.5 | 12.3/12.1 (11.7/11.4) |
| |||||
For 34-protein validation dataset | |||||
| |||||
TALOS+ | 84.7% d | 4.2% e | 15.3% d | 12.5/12.0 | 13.7/12.7 |
TALOS-N | 91.4% d [87.5%/3.9%] | 3.5% e [2.8%/23%] | 8.6% d | 8.5/8.3 | 12.2/12.1 (11.7/11.5) |
TALOS+ and TALOS-N predictions were performed for the 580-protein chemical shift database and for a separate 34-protein validation dataset.
Average standard deviation of φ/ψ torsion angles among the 10 or 25 best matched tripeptides/heptapeptides for “Good” TALOS+/TALOS-N predictions, representing the average precision of the predictions. See footnote e for the definition of a good/bad prediction.
Rmsd values between TALOS+/TALOS-N predicted φ/ψ angles (“Good” predictions only) and φ/ψ angles observed in the reference structures, representing the average accuracy of the predictions; values in parenthesis correspond to residues for which both both TALOS+ and TALOS-N reported “Good” predictions.
Fraction of “consistently” predicted residues relative to the total number of residues for which predictions are calculated. Percentages for the “Strong” and “Generous” TALOS-N predictions are given in square brackets.
Percentage relative to the total number of “consistently” predicted residues (“Good”+”Bad”). A “bad” prediction is defined based on the criterion sqrt[(φobs − φpred)2 + (ψobs − ψpred) 2] > 60°; otherwise the prediction is defined as “good”. Percentages for the “Strong” and “Generous” TALOS-N predictions are given in square brackets.