The dynamic programming algorithm of Myers and Griffiths (2003) [30] implemented in the PAIRWISE program was used to identify a list of all pairs of sites with evidence of recombination. The positions of these pairs of sites in the DENV genome were used to determine if they are localized within codons (intracodon). Coalescent simulation of codon sequences The codon sequences of dengue virus serotypes were simulated by the coalescent method of Arenas and Posada (2010) [20]. It is based on the coalescent with recombination method under a Wright-Fisher

neutral model [31]. The ‘Netcodon’ algorithm developed by Arenas and Posada (2010) [20] was used to simulate DENV codon sequences with serotype specific recombination Barasertib concentration rates estimated by PAIRWISE and the M1 codon model. This codon model incorporates two categories (ω0 P0, ω1 P1) of values to represent proportions (P0 or P1) of non-Ro 61-8048 in vitro synonymous to synonymous substitutions (ω0 or ω1) in the sample sequences. The other parameters such as mutation rate, nucleotide frequency of coding sequences, transition/transversion ratio estimated from the observed data by DnaSP [23] were used in generating simulated data sequences. The simulation MM-102 was carried out to generate 10 replicates of 65 samples, which generated 650 random sequences of the DENV coding genome. The simulated

data were then analyzed by PAIRWISE to identify all the pair-wise sites showing evidence of recombination and to determine if they are localized

within codons (intracodon). Statistical analysis All statistical analyses were performed in R. The 2×2 contingency tests were conducted either by Yeats’s Chi square tests or by Fisher’s Exact tests depending upon Protein kinase N1 the sample sizes. All p-values are two-tailed. Statistical significance of association between intracodon recombination and purifying selection was measured by hypergeometric tests as per method described in Fury et al. (2006) [32]. Briefly, the distribution of sites of purifying selection (n1) and the sites showing intracodon recombination (n2) among all the recombination sites (n, which are identified from PAIRWISE analysis) were determined. The total number of possible choices for the two groups of sites was calculated as C(n, n1)* C(n, n2). Similarly, the total number of possibilities for choosing the purifying sites was C(n, n1), whereas the number of possibilities for choosing the purifying sites showing evidence of intracodon recombination was C(n1, m), where m is the total counts of sites showing evidence of both purifying selection and recombination within codons. Among the total number of sites in the genome identified as sites with intracodon recombination, the remaining n2-m sites were chosen among the remaining n-n1 purifying sites in C(n − n1, n2 − m) ways.