Fig. 3: Comparison, development and validation of the iCipa model.
A Previous models of coiled-coils predict the interactions in the CCNG1 with low R2. The black line represents a linear model of the interaction scores predicted by different algorithms. B Coefficient of determination of interaction scores of different iCipa candidates evaluated during development. Each point represents one bootstrap of the data. N = 100 bootstraps. Boxplot center lines represent the median, the hinges represent the 25th and 75th percentiles and whiskers represent the largest/smallest value within 1.5x it’s respective hinge. ***p < 10−15 by two-tailed t-test. C iCipa is more predictive of interaction scores (R2 > 0.27) than the previous models shown in (A). Black line represents a linear model of interactions scores, as predicted by iCipa scores. D Weights for the iCipa model. Each weight scores a pair of amino acid residues at specific registers between the coiled coils (at a-a’: NN and II, at e-g’ and g-e: KE, KK and EE). E iCipa is more predictive of the previously published CC0 melting points than the bCipa or Potapov scoring functions. Individual dots represent melting points as compared with the normalised score from one of the three scoring algorithms. Boxplot center lines represent the median, the hinges represent the 25th and 75th percentiles and whiskers represent the largest/smallest value within 1.5x it’s respective hinge. Source data are provided as a Source Data file.
