2.6. Codon Pair Bias Analysis

Similarly, but independently of codon bias, the juxtaposition of codons in ORFs does not appear to be randomly distributed either [32], and these preferences are typically referred to as codon pair bias. An algorithm has been developed to quantify codon pair bias [33]. For each of the 3721 possible codon pairs, excluding stop codon pairs, the codon pair score (CPS) is defined as the natural log of the observed ratio over the expected number of occurrences of each codon pair’s overall coding regions. It can be calculated by using the following equation [33]:

where the codon pair AB encodes for amino acid pair XY, and F denotes frequency (number of occurrences). The CPS value for a given pair determines whether the pair is over-represented (+) or under-represented (−). With the calculated CPSs, we can further calculate the codon pair bias (CPB) score as follows [33]:

where k indicates how many kinds of codon pairs there are. The CPB score has already been used for virus attenuation through deoptimization [33,34,35]. Theoretically, a decreased CPB score is associated with the inefficiency of the viral gene translation in the host, which results in attenuation of viral replication [36]. The CPB scores were computed by the CPBias package (version 1.0) (https://github.com/alex-sbu/CPBias/) (accessed on 5 January 2021) of R.