RNA genome conservation and secondary structure in SARS-CoV-2 and SARS-related viruses

Ramya Rangan; Ivan N. Zheludev; Rhiju Das

We sought to further check structured windows reported by RNAz using orthogonal approaches. First, we explored using R-scape to make structure predictions with covariation signal in the sequence alignments. However, we found that the SARSr-MSA-1 alignment had insufficient variation to detect conserved base pairs with covariation, lacking alignment power for all genomic windows. 22 Second, we validated structured window predictions with alifoldz, a program that calculates a zscore for an alignment window by comparing the window's consensus minimum free energy structure to that of random shuffled alignments. To mirror the RNAz analysis above, we scanned through windows of length 120 nucleotides sliding by 40 nucleotides. We chose a z-score cutoff of -2.69, which kept only 1% of windows when running alifoldz on all shuffled windows across the genome. This approach led to predicting 228 alifoldz structured windows, overlapping with 104 of the 117 RNAz structured windows (P>0.9 cutoff). This overlap is statistically significant (p-value<1e-05). RNAz structured windows supported by alifoldz analysis are highlighted in Supplementary File 1.