The individual UTR RNA segments and the NS protein segments from each full-length genome sequence were retrieved and then connected to create new sequence components for covariation analysis. These 6 binary sequence components were input to the Weka software to determine the covariation association between each of the nucleotide sites and the amino acid residues. The unique association rules of these binary sequence datasets are summarized in Table S2. Thirty-nine unique association rules were identified. Results in the set for all genotypes indicate covariance of the 204th nucleotide of the 59UTR with 3 amino acid residues of the NS3 protein and the 243rd nucleotide of the 59UTR with 6 amino acid residues of the NS2 protein and 3 amino acid residues of the NS3 protein . Since the covariance between 59UTR243 and NS2-14, -41, -76, -110, -211, -212 and NS3-71, -175 and -621 consists of associations involving the largest number of multiple sites, the functional relevance of 59UTR243 in co-variation with the residues in the NS2 and NS3 proteins but not the other pairings was examined in our cell-based experiments. ovariations were WZ4002 biological activity introduced in order to analyze their effects on the replication efficiency using a transient-replication assay. We constructed 9 pairs of variants in the context of the wild-type NS2-39 replicon , each consisting of a single amino acid substitution at the NS2 or NS3 region and double substitutions in combination with 59UTR-G243A and the corresponding amino acid . Based on the normalized luciferase activities at 3 consecutive time points, the transient luciferase assays order PLX4032 indicated that the 9 single amino acid variants decreased replication efficiency in the presence of 59UTR243G, but replication efficiency could be rescued when any single variant of NS2-I41L, NS2-I76V, NS2- I110L, NS2-G211S, NS3-I71V and NS3-M175L was combined with 59UTR-G243A. On the contrary, the 59UTR-G243A could not compensate the NS2-F14L, NS2-Q212K and NS3-A621T variants. Furthermore, different types of codon usage were introduced for NS2-I110L and NS2-G211S , yielding comparable compensatory effects and indicating that differences of codon usage at the nucleotide level may not be a concern . These results together suggest that the covariation of 59UTR-G243A with the NS2 and NS3 proteins was most likely due to amino acid substitution, but this was not the case for the specific nucleotide sequences. Data mining involves finding patterns or rules in large data sets.