Figure 3.

The protein and mRNA abundances, the ribosome density and the protein length can be predicted from the sequences. We used the same models as in Fig. 2 to predict the protein and mRNA abundances, the ribosome density and the protein length. As in Fig. 2, the models were based on (1) the protein composition – the amino acid and codon percentages; (2) these percentages and features derived from the sequence; (3) all of the previous features and additional overall features. The bar graphs show the Pearson’s correlation coefficients between the measured protein and mRNA abundances, ribosome densities and lengths, and the respective values predicted by our models, as in Fig. 2d. Remarkably, the amino acid and codon compositions alone are sufficient to predict these parameters with an accuracy that is on average ~70% of the maximum expected (the reproducibility of the data between different data sets, from different laboratories).