Variation at Spike position 142 in SARS-CoV-2 Delta genomes is a technical artifact caused by dropout of a sequencing amplicon

Abstract

Public SARS-CoV-2 genomes from the Delta lineage show complex and confusing patterns of mutations at Spike codon 142, and at another nearby position, Spike codon 95. It has been hypothesised that these represent recurrent mutations with interesting evolutionary dynamics, and that these mutations may affect viral load. Here we show that these patterns, and the relationship with viral load, are artifacts of sequencing difficulties in this region of the Delta genome caused be a deletion in the binding site for the 72_RIGHT primer of the ARTIC V3 schema. Spike G142D should be considered a lineage-defining mutation of Delta.

Keywords: artic; delta; genome.

Figure 1.. G142D is fixed in Delta, with almost all Delta sequences where the nucleotide at position 21987 has an A at this position.

In contrast, Alpha contains the reference G at this position. 35% of Delta sequences have N at this position, indicating that the position does not have sequencing coverage.

Figure 2.. Local sequencing coverage (note log Y axis) near amplicon 72.

Lines along the top show amplicons 71, 72 (blue) and 73, as well as parts of amplicons 70 and 74. Dashed portions are primers (so ought not to be present in the final sequence). In the top panel, the red line is average coverage for 10 random B.1.617.2 sequences with D at spike position 142; grey line average coverage for 10 random B.1.617.2 sequences with N (missing data) at that position. Bumps in coverage are caused by overlap of adjacent amplicons and overlap of paired-end illumina sequence reads in the middle of each amplicon. The B.1.617.2 lineage defining deletion at 22029–22034 causes the zero coverage dip in the right-hand primer for amplicon 72. In the bottom panel, Alpha genomes (blue-dashed) have good coverage throughout, and show dips at two characteristic deletions. The blue-solid line shows coverage for the small number of sequences typed as Delta with G at spike 142, which appear to be mixtures of Delta and Alpha.

Figure 3.. 72_RIGHT primer of the ARTIC V3 scheme shown against the SARS-CoV-2 reference genome, with the lineage defining deletion at 22029–22034 in Delta highlighted.

Figure 4.. Relationship of the residue at 21987 in Delta lineage samples (representing spike 142) with the residue at 21767 representing spike 69/70del for the set of UK samples.

Most of the small number of "revertants" with G at position 21987 also have a gap at 21767, corresponding to the spike 79/70 deletion found in B.1.1.7. This suggests that these sequences represent contamination from B.1.1.7 (either as a mixed infection or in the laboratory) that is specific to amplicon 72 because of the reduced efficiency of this amplicon in Delta samples.

Figure 5.. G residues at position 21987 in Delta result in large part from reads which appear to have come from the 73_LEFT primer.

This plot shows seven samples randomly selected from the SRA with a G at position 21987 of the consensus sequence. In all cases there is substantial evidence for A at this position. Specifically reads which must have come from amplicon 72 (indicated as <= 29160) show substantial evidence for A, whereas reads likely to have come from the 73_LEFT primer sequence (indicated as > 29160) are overwhelmingly G. (Panel A features some jitter to reduce overplotting.)

Figure 6.. Distribution of nucleotides at 21846 (nucleotide T encodes T95I) for different sublineages within Delta.

T95I is fixed in AY.4, but absent from other currently designated Delta sublineages.

Figure 7.. Relationship of C _t value and residue at position 21987 for COG-UK Delta samples until 30 June 2021.