Conservative and non-conservative amino acid substitutions.

Conservative and non-conservative substitutions used for the computation of λ. Only amino acid mutations that are direct neighbors in the genetic code are considered in this analysis. A less stringend definition where more substitutions are classified as non-conservative despite being less often predicted to be destabilizing, e.g. to validate that this classification does not introduce systematic amino acid biases, preserves all observed trends, albeit with less statistical significance.