Detect Positive Selection in Type IIL Restriction Enzyme Using PAML

The role of positive selection in molecular evolution has long been debated since the proposal of the neutral theory. With the explosive growth of genomic sequence data and aided by statistical methods such as Phylogenetic Analysis by Maximum Likelihood (PAML), recent years witnessed a flurry of reports of positive selection in protein-coding genes. However, with a few exceptions, the reported positively selected sites generally lacked functional data to support them. Restriction enzymes in bacteria are under constant selective pressure from the bacteria-phage arms race, and in some enzymes amino acids responsible for the specificity changes are known. Here I tested the performance of PAML using the functionally well characterized type IIL restriction enzymes as the model system. The result showed that PAML was highly consistent in detecting positive selection in the sequence data. However, in terms of identifying positively selected sites, PAML was very sensitive to the sampling bias in the dataset and had a high false negative rate and also possibly a high false positive rate. The result suggests positively selected sites identified by PAML should be treated with caution and validated by functional studies.