Time series analysis of SARS-CoV-2 genomes and correlations among highly prevalent mutations

The efforts of the scientific community to tame the recent SARS-CoV-2 pandemic seems to have been diluted by the emergence of new viral strains. Therefore, it becomes imperative to study and understand the effect of mutations on viral evolution, fitness and pathogenesis. In this regard, we performed a time-series analysis on 59541 SARS-CoV-2 genomic sequences from around the world. These 59541 genomes were grouped according to the months (January 2020-March 2021) based on the collection date. Meta-analysis of this data led us to identify highly significant mutations in viral genomes. Correlation and Hierarchical Clustering of the highly significant mutations led us to the identification of sixteen mutation pairs that were correlated with each other and were present in >30% of the genomes under study. Among these mutation pairs, some of the mutations have been shown to contribute towards the viral replication and fitness suggesting the possible role of other unexplored mutations in viral evolution and pathogenesis. Additionally, we employed various computational tools to investigate the effects of T85I, P323L, and Q57H mutations in Non-structural protein 2 (Nsp2), RNA-dependent RNA polymerase (RdRp) and Open reading frame 3a (ORF3a) respectively. Results show that T85I in Nsp2 and Q57H in ORF3a mutations are deleterious and destabilize the parent protein whereas P323L in RdRp is neutral and has a stabilizing effect. The normalized linear mutual information (nLMI) calculations revealed the significant residue correlation in Nsp2 and ORF3a in contrast to reduce correlation in RdRp protein.