Correlation between Ka/Ks and Ks is related to substitution model and evolutionary lineage

J Mol Evol. 2009 Apr;68(4):414-23. doi: 10.1007/s00239-009-9222-9. Epub 2009 Mar 24.


In 2005, Wyckoff and coworkers described a surprisingly strong correlation between Ka/Ks and Ks in several data sets using the LPB93 algorithm. This finding indicated the possibility of a paradigm shift in the way selection strength can be measured using the Ka/Ks ratio. We carried out a calculation of Ka and Ks using six different algorithms on three cross-species orthologous data sets and found a highly variable correlation among the algorithms and lineages. Algorithms based on the GY-HKY substitution model exhibit a weaker positive correlation or a stronger negative correlation than those based on the K2P and JC69 substitution model. Even if one algorithm shows a positive correlation between Ka/Ks and Ks in a warm-blooded lineage, it may show no correlation in a cold-blooded lineage. This algorithm-related and evolutionary lineage-related correlation indicates the need for great caution in drawing conclusions when using only one Ka and Ks algorithm in a genomewide analysis of selection strength. Our results indicated that currently used algorithms for Ka and Ks calculations are flawed and need improvements.

Publication types

  • Research Support, Non-U.S. Gov't

MeSH terms

  • Algorithms
  • Animals
  • Computer Simulation
  • DNA Mutational Analysis
  • Data Interpretation, Statistical
  • Evolution, Molecular*
  • Humans
  • Mice
  • Models, Genetic*
  • Polymorphism, Genetic*
  • Rats
  • Sequence Alignment
  • Sequence Analysis, DNA