Thread by @Kevin_McKernan

Something @lifebiomedguru alerted me to many moons ago.
There is another 1252 amino acid Open Reading Frame across the spike mRNA. Given the mRNA is all Crick strand this should never exist in the vaccine unless the plasmid is there providing the Watson strand that codes for it

It does not have Kozak consensus (signal for making proteins) sequence or a promoter (signal for making mRNA).
However Internal Ribosomal Entry Sites (IRES) are not easy to predict computationally. They must be experimentally determined.
T7 Polymerase is known to strand flip.

This means while its making mRNA on the crick strand, if it runs into a hairpin in the RNA, it can flip strands and start copying the other direction. Its never helpful to have the reverse strand be a full in frame ORF. Most codon optimizations prevent this.

In codon optimizations, You have the freedom to change the wobble bases to make stop codons when read in reverse. They didn't do this.
jbc.org/article/S0021-…

The EMA had a lot of questions regarding the dsRNA contamination. Pfizer is measuring this with an ELISA assay that might be blind to N1-methyl pseudouridine.

https://t.co/99kEjR6sR4nature.com/articles/s4159…

An artifact we witnessed in RNA-Seq.

When you sequence mRNA it should all be Crick strand (Sense strand). The opposite strand is called the Watson strand (Anti-sense). In this case both the Watson and Crick strands are sense due to the reverse ORF.
You can see this

when we sequence RNA. The T7 promoter stick out like a sore thumb(2100). That where the mRNA is being expressed from. The material before it has equal ratio of Watson/Crick sequence. Thats the DNA plasmid backbone. What you should see after the T7 promoter is all Crick strand.

We don't see that. We see areas of more Watson than Crick sequence. Thats dsRNA contamination.

We need other tools to quantitate this.
We could tile 30 qPCR amplicons across the target and measure before and after treatment with RNase iF.
bmcbiotechnol.biomedcentral.com/articles/10.11…

RNase If only chews single stranded RNA. leaving dsRNA behind.
Or use the read counts and strandedness in RNA-Seq to estimate it. This is likely the most informative approach.
30 qPCR assays would have to be SYBR green qPCR to keep the costs down.

The codon optimization did change the hairpin potential. N1-methyl pseudoU likely makes these more stable.

Not certain what this means other than it is an avoidable problem in codon optimization and 1252 amino acids coding on both strands is a hell of a feat.
This is only seen in BNT162b2. Haven't looked in their Omicron sequence yet.

The strandedness map above is for Moderna. The Pfizer RNA-Seq created a truncated sequence (7479bp) compared to the sequencing we did using RNase to get rid of the RNA (7810bp). Asimil ar pattern emerges with Pfizer. In this case the plasmid T7 promoter is running Right to left