Thread by @MackayIM

That website opinion attack piece on the Corman/Drosten RT-rPCR tests has a few confused & sometimes very wrong comments.
I'll preface this thread with a *presumption* I'm making - these authors haven't worked on or with this test themselves. And that matters a lot if true.

I'm not going through everything but here's one from their "Top10 things I hate about this test" list.
From Number 5. "nor does it contain any other negative controls"
But Corman et al. are actually very clear that negative controls were included.

"all assays were tested 120 times in parallel with water and no other nucleic acid except the provided oligonucleotides. In none of these reactions was any positive signal detected"

But even better, they tested primer specificity on a lot of human viruses, including CoVs:

The outcome was good.
"Additional undiluted (but not quantified) cell culture supernatants were tested as summarised in Table 2. These were additionally mixed into negative human sputum samples. None of the tested viruses or virus preparations showed reactivity with any assay"

These cover all the types of viruses we'd expect to cause problems if since they may also be found in man samples at the same time/season. But they *didn't* cause problems & this is because they are genetically distinct viruses (yes, even the human CoVs!). PCR is very specific.

Next, I'll cover the "minor concerns" that the "INTERNATIONAL CONSORTIUM OF SCIENTISTS IN LIFE SCIENCES (ICSLS)" who put together the Corman-Drosten attack piece, had with the Eurosurveillance paper.
These are not about the science or test about editing text.

These were issues with Tables and Figures & are actually problems with copyediting (ie the journal missed them). Yes, scientists make errors in writing (like other writers😂), usually caught by the journal
1. nanomolar was sometimes incorrectly abbreviated as nm instead of nM.

💨Scientists in this area kinda know what this should be used & it wouldn't have had a single one rushing to reconstitute their oligos with a length (instead of a volume) of water. Easy typ0 for @Eurosurveillanc to correct.

2a. The Consortium declared "highly unusual to do [sic] alignment with reverse complementary writing of the primer sequence as the authors did in figure 2"
💨As stated in Fig title, it depicts oligo "binding regions". It shows where they land. For 5'-3' sequence, see Table 1

2b. Consortium: "in addition, a wobble base is marked as “y” without description of the bases the Y stands for.
💨Yep. Missed by reviewers & journal but in the actual primer sequence one would order, it's an R (because the landing site is the reverse complement of the Rev primer)

Point 3. Consortium say it's misleading to not have Tms and GC values in Table 1.
💨Both are easily calculated by the user. However, I *never*, rely on the lab's published Tm. It's a guide at best. If I'm trying someone's oligos, then I'm trying them in my house.

That means they will need optimisation on my cyclers, in my mix and in my hands. Similarly GC content is part of the design process. You care if you are designing your own (because you cntrl these variables then), but if you've decided to try someone else's, then it is what it is

So on these minor points - I'd agree. They are bloody minor.
At best they are an oversight of the journal's editing process and at worst, the authors could write to the journal asking for these to be corrected.

NEXT is the Consortium's major concern No.1 which revolves around primer and probe (which I'll call oligonucleotides, or "oligos) design.
It's a multipart point.
1a. Concentrations are "erroneous".
💨No. Simple. Experts in use of PCR as an in vitro diagnostic test know that

it's really important to test a range of oligo concentration to find those which produce the best curve shape and earliest Ct values. We can do this with what's called a chequerboard titration. This tests a range of oligo concentrations, in every permutation & combination.

An example shows concentration matters. If higher concentrations work better, that's what you use. People unfamiliar with PCR for quality detection of pathogens in human samples may not do this, academia in particular, but its really important. And concs can vary for each test.

Once optimal oligo mix is settled on, the validation process that follows will identify if there are any strange results. The concerns of 800nM or 900nM being too high is mostly a carryover from pre-real-time PCR (rPCR) days. It looks like Corman-Drosten did do this workup 👍

Point 1b. The use of degenerate positions in oligos affects reliability. A degenerate position is a nucleotide position in an oligo that could be any of 1, or 2 or 3 or 4 of the nucleotides (A,C,G or T or sometimes a special base). We use a code when ordering the oligo. For e.g.

we add a special code when we *order* the oligo, which tells the company to make a mix of oligos with each of the nucleotides at that position. See codes at bioinformatics.org/sms/iupac.html
If we ordered an oligo ACYTG we'd receive, in a tube, a mix of ACcTG and ACtTG oligos. This is

used when we know or suspect that at a certain position, there might be viral variants, some of which have one or other (or other or other) sequence at that position. It's not unusual & any are issues are again watched for during optimisation and validation.

1c. GC content-talked about👆
1d. Not detecting whole genome & secondary structure,
💨This entire section has some very loose language. Not what I'd expect from experts who are fluent in eth lingo of this field.
The suggestion that we should amplify the entire genome by RT-rPCR

("terminal primers on both ends of the viral genome") is ridiculous.
💨That is for whole-genome sequencing purposes which remains too slow for the sort of turnaround times we need from a pathology lab.

The idea that 3 distinct "primers" (tests?) need to be made to identify SARS-CoV-2 comes from, nowhere.
💨Two distinct targets would be ideal. More isn't necessary but of course, if you have time & resources and sample, sure. But not essential.

The Consortium is also concerned that the entire genome is not being detected by 2 or 3 RT-rPCR tests.
💨It doesn't have to be for an RT-rPCR to identify the presence of a virus in a person when that presence marks an abnormal presence (=infection). No RT-rPCR does this

On language: "This is why the RT-PCR Test used in clinical routine for detection of COVID-19 is not indicated for COVID-19 diagnosis on a regulatory basis." -
💨We don't detect COVID-19 with RT-PCR, we detect SARS-CoV-2. This highlights, again, a lack of basic understanding.

The Consortium identified a match for one primer with another pathogen but not the probe. The other primer's nearest match was very distant (400k; & in the wrong orientation).
💨a non-starter. It takes 3 oligos to tango in an RT-rPCR

The Consortium was worried by 2 oligos having some matching sequence (the renamed Corman_E_Sarbeco_R & the renamed Corman_RdRp_SARSr-F1, but that's a hangover from the preprint version-I presume that they mean RdRp_SARSr-F; hey, they chose to be pedantic, not me).

💨they haven't provided the settings they used for this predictive calculator (thermofisher.com/au/en/home/bra…), but using the default settings, I was unable to replicate their result (always good to include a detailed methods section so a study can be replicated elsewhere when writing a

real scientific manuscript; or else it may be taken for a pseudoscientific amateur endeavour & be dismissed outright.
M
Dimers and secondary structures happen sometimes. You may have a target region that is optimal so you can't move your primers. Then you *have* to just order

your best designs for that region & see how the optimization and validation pan out.
Sometimes what looks good/bad in silico (in the computer) actually works horribly/perfectly, respectively, when you test the RT-rPCR.

It's worth pointing out that Corman-Drosten don't recommend duplexing (combining) these 2 different tests so these primers clearly should never be in the same tube unless someone else decides to do that. Corman-Drosten identify a *workflow* of successive test use.

I hope you get a sense in this thread that the importance of understanding diagnostic molecular virology is really key to using "the PCR" for human sample analysis. Quality among the trained and experienced staff & the processes is essential.

The Consortium also claimed that "These are severe design errors, since the test cannot discriminate between the whole virus and viral fragments. The test cannot be used as a diagnostic for SARS-viruses."
💨They aren't errors at all. This is a weird claim that wrongly expects a

rapid diagnostic virology RT-rPCR to be able to amplify an entire 30,000 nucleotide RNA genome. This isn't their purpose. Also, this is thermodynamically ...let's say unlikely... in 45 cycles of 95°C/15sec, 58°C/30sec preceded by a 10min RT step.
We have whole-genome sequencing

to address this need. It's currently a much more complex & lengthy process with a low throughput unsuited to identifying virus in many thousands of samples per day, per lab (that may change in the future of course).

A side-note. There seems to be an implication throughout the Consortium's writings that the world is only using Corman-Drosten's tests.
In reality, shortly after their test appeared on @WHO site in January, it was followed by tests from China, Japan, Thailand, HK, France & US

who.int/publications/m…
Other tests were developed & not put on the WHO site (I was involved in a few) - some of which have & will keep appearing in scientific publications.

Each is its own distinct PCR test that targets its own stretch(es) of sequence; although some may overlap in their oligo landing sites.
I can't speak to how well others' optimized & validated their tests.
Then there are commercial kits-it's a bit harder to know what's in them.

But I can say that the Corman-Drosten lab are not new to making good RT-rPCR tests. I first came across this in the field of MERS-CoV diagnostics where they were equally rapid, generously collaborative & thorough. The Consortium fails to see that these tests

helped some countries to respond quickly & detect cases early. Perhaps a review of what those countries did next could be written instead of one this one paper & this 1 research group with whom at least some of the Consortium authors appear to have an unhealthy fixation

The recently added "Addendum" seeks to bolster the Consortium's lack of relevant experience with examples that reportedly prove all of its "Review" points..
Reference No.1 (citation No. 16) is a Letter by Pillonel et al.

They pointed out the reverse primer of Corman et al's RdRp test had a degenerate base which didn't match later SARS-CoV-2 sequences. The original sequence used an S [=a C or G at that position]; Pillonel suggested an R would be better [=A or G] for SARS-CoV-2 (agree but I'd

argue that a V would have been best for the *original* purpose).
But as Corman & Drosten wrote in their reply to this letter, these tests were designed *before* there was virus to test them on in Germany. They were designed to catch the virus based on an expected relationship to

SARS-related CoVs but were catering for genetic variation. The S would have been perfect to bind to all other SARS-rCoVs. They bet wrong on this base because it ended up being different from SARS-rCoVs. Still a good approach to accommodate *potential* variation among variant

C&D stated that this didn't affect the test's sensitivity in their own experiments. It's nonetheless clear from Nalla et al, that this was the least sensitive of the 3 tests from Corman et al, for whatever reasons.

So this example does highlight an issue but as we've come to expect from the Consortium's writings, they made no effort to put the context around this point. Hopefully, you have that now.

Refs in that list:
eurosurveillance.org/content/10.280…
eurosurveillance.org/content/10.280…
jcm.asm.org/content/58/6/e…

The Nalla et al article (Ref 24 in the Consortium's addendum; jcm.asm.org/content/58/6/e…) notes that the Croman et al E gene test was in the top 2 for sensitivity "with no false-positive results observed".
Quite a different conclusion from the attack piece's overall narrative.

I'm just going to stop for a tick & go back.
The WHO now notes on their Molecular Assay introductory page that the early tests are not being updated.
Always good when using someone else's old design to check the oligos.
who.int/publications/m…

Addendum Section 1.A.b.1,. Muenchhoff et al. (Ref No.1). The paper includes Drosten as a co-author.
eurosurveillance.org/content/10.280…
The Consortium use this paper to query COIs, the need for such speed & that the RdRp test "must be exchanged and/or removed from the protocol"

First up, it simply concludes "the protocol needs to be amended to improve the sensitivity of the RdRp reaction". As you'll see below, that's now done & in print.
Let's not add emotion & words where none exist.
The @WHO site hasn't updated-they should add more warning.

COIs - the Consortium note that a lab affiliation is missing & I think they are implying a commercial gain for Drosten to be had by publishing these primers? Not sure how that works or their point. Usually, commercial kits don't describe oligo sequences used.

That's' not really about this paper supporting their hypothesis though.
Muenchhoff et al also get to the bottom of the RdRp assay being less sensitive...

The authors modified the reverse primer, fixing that degeneracy to make it more SARS-CoV-2, instead of SARS-rCoV,-specific & moving it to raise the Tm. The figure below is a great example of how variable PCR results are between labs (look at the different kits & instruments!)

Just to add, that the change to the primer sequence might have rendered it less useful for detection of SARS-rCoV as it included a variable position at one the newly added primer nucleotides (see more on this from the previous paper thread).

These change improved RdRp test sensitivity, but it still lagged behind some other tests-that's how this goes sometimes.
Other tests that are worse though. I guess the Consortium is in the process of writing "reviews" for those too? If not, is this a Drosten et al attack?

Muenchhoff et al showed the Seegene E commercial kit test performed with less sensitivity than the modified RdRp test.

Lastly on this paper, the Consortium asks the question: "Why the race to get these primers to testing companies and onto the WHO website in January 2020?"
I might leave that to you, dear reader, to think through. Why would having testing in place early be useful for saving lives?

So this paper updates the 1st gen RdRp test & gives those using it another option to confirm positives. The biggest takeaway for me is hosts of early test need to ensure they are either updated to latest versions or else have big red warnings all over the page & the PDF

And it showed that this modified version of the RdRp test is more sensitive than the US CDC N1 or Corman et al E gene test.
Win!