Fig. 4

Non-canonical junctions are not associated with TRS-like homology. a Illustration of computational approach used to assess homologous sequences between the 5′ and 3′ junction points. For each of the three dRNAseq datasets, the 30 bases flanking the 5′ and 3′ junction points were assessed, and the longest homologous sequence between these two 30 base regions was determined. b–d Homologous sequences present in the 15 bases on either side of the 5′ and 3′ ends of each junction. Junctions are classified by the location of their 3′ end—if this is within 15 bases of the canonical TRS site or if it falls within a gene it is assigned accordingly. The only exception is ORF1a—junctions with a 5′ end originating in ORF1a are assigned to ORF1a. Labels represent the most common homologous sequence between the ends of each junction. The core TRS sequence ACGAAC is underlined. e–g The length of the longest homologous sequence for canonical (C) and non-canonical (NC) are plotted. Here, junctions were considered canonical if they have a 5′ end within 20 nucleotides of the TRS-L and a 3′ end within 15 nucleotides of a TRS-B. (R) represents random homology lengths. The points for (R) were calculated by first extracting all possible 30 base sequences (30mers) from the SARS-CoV-2 genome, and then assessing the length of the longest homologous sequence between 100,000 random pairs of 30mers that are separated by at least 1000 bases on the genome. Within each column, the relative width of each band represents the relative abundance of junctions with each homology length. The value at the top of each column is the mean homology length