Argues SARS-CoV-2’s genome is consistent with laboratory origin via serial passage or targeted recombination: proposes the virus is chimeric, with a RaTG13-like backbone and an RBD nearly identical to a pangolin coronavirus (MP789), plus the unexplained furin cleavage site; argues this combination is a priori improbable via a single natural recombination event and would have been straightforward to construct with 2018-era reverse-genetics techniques. relevance_note: the central published articulation of “the genome looks chimeric/engineered,” and the paper that triggered the formal BioEssays rebuttal exchange with Tyshkovskiy & Panchin.
Observations
O-74 - SARS-CoV-2 has a RaTG13-like backbone but a pangolin-CoV-like RBD (chimeric structure)
SARS-CoV-2 is ~96.2% identical genome-wide to bat CoV RaTG13 (its closest full-genome relative) and 79% to SARS-CoV, yet its RBD is nearly identical to the pangolin coronavirus MP789: SARS-CoV-2 and pangolin CoV share identical amino acids at the five critical RBD contact residues while RaTG13 shares only one - a chimeric backbone-vs-RBD split.
Link to original
O-75 - The furin site is a 12-nt PRRA insert with a rare doubled CGG-CGG arginine codon and a FauI site
The SARS-CoV-2 furin cleavage site is created by a 12-nucleotide insert (TCCTCGGCGGGC) coding PRRA at the S1/S2 junction; the two adjacent arginines are encoded by CGG-CGG, a rare codon in these viruses (only ~5% of arginines use CGG, and this is the only doubled CGG in the genome), and the insert contains a FauI restriction site.
Link to original
O-76 - The PRRA furin-site insertion is out of frame relative to aligned RaTG13-MP789
Aligned against RaTG13 and pangolin MP789, the PRRA insertion is out of reading frame: it splits the original serine codon TCA into a new serine codon (TCT) plus part of an alanine codon (GCA).
Link to original
O-77 - No close relative has a true S1-S2 insertion; RmYN02's reported insertion is only substitutions
The polybasic furin site and its PRRA insert are absent from all related lineage B betacoronaviruses; the RmYN02 virus reported to carry a natural PAA insertion at S1/S2 shows, on nucleotide alignment to its closest relatives (bat-SL-CoVZC45/ZXC21), only substitutions and no actual insertion.
Link to original
Hypotheses
H-39 - SARS-CoV-2 is a laboratory-engineered chimera (RaTG13-like backbone + pangolin-like RBD + site-directed furin site)
SARS-CoV-2 was created in the lab by combining a RaTG13-like bat backbone with a pangolin-CoV-like RBD and inserting the furin cleavage site by site-directed mutagenesis, producing a chimera that would appear natural.
Link to original
H-40 - SARS-CoV-2's RBD-hACE2 fit was perfected by serial passage in cell culture or humanized-ACE2 animals
The optimal adaptation of SARS-CoV-2’s RBD to human ACE2 (and possibly the O-linked glycans) could have been achieved by serial passage of a chimeric or bat virus in cell culture or in humanized/hACE2-expressing animals in the lab.
Link to original
Arguments
A-68 - A single natural recombination producing the backbone+RBD+FCS combination is a priori improbable
Producing SARS-CoV-2’s RaTG13-like backbone with a pangolin-like RBD by natural recombination would require a RaTG13-like and a pangolin-like virus co-infecting one cell in one host simultaneously - improbable given low pangolin population density, the scarcity of these CoVs in pangolins, and that reconstituted RaTG13 does not bind pangolin ACE2 - so an engineered chimera is a competitive explanation.
Reasoning
Natural recombination requires two divergent parental viruses to be present in the same cell of the same animal at the same time. For a RaTG13-like backbone to recombine with a pangolin-CoV RBD, a host (e.g. a pangolin) would need co-infection by both lineages. The authors argue this is unlikely because pangolins occur at low population density and rarely carry these coronaviruses, and - critically - receptor-binding studies show reconstituted RaTG13 does not bind pangolin ACE2, so a RaTG13-like virus is unlikely to have productively infected a pangolin to meet the pangolin CoV in the first place. If the single natural-recombination route is improbable, then deliberate laboratory recombination - trivial to perform and leaving the same chimeric signature - becomes a competitive explanation for the observed backbone/RBD split.
Validity verdict — approved (checked)
Reconstruction. Premises: (i) natural recombination requires two divergent parents co-infecting one cell of one host; (ii) for a RaTG13-like backbone + pangolin-CoV RBD this venue is a pangolin, which is low-density and rarely carries these CoVs; (iii) reconstituted RaTG13 does not bind pangolin ACE2, so a RaTG13-like virus is unlikely to have productively infected a pangolin; (iv) lab recombination is trivial and leaves the same chimeric signature. Load-bearing step: from “the pangolin-mediated natural recombination route is improbable” to “an engineered chimera is a competitive explanation.”
Evaluation. This is a comparative-likelihood move: if P(observed chimeric structure | natural) is depressed while P(observed | engineered) is not, the likelihood ratio lifts the lab hypothesis to “competitive.” Conditional on the premises the step is valid, and the conclusion is suitably weak (“competitive,” not “confirmed”), so no completion-by-force. Note the argument carries an implicit premise that the recombination venue must be a pangolin (the receptor argument only bites for pangolin as host); a recombination in a third host, or the finding that the RBD actually nests in the wild BANAL bat clade (see A-66), would undercut premise (ii)-(iii). But those bear on the truth of the “natural route is improbable” premise, priced at step 8 — not on the inferential step, which holds given the premises. Approved; traced here → checked.
Link to original
A-69 - The out-of-frame furin-site insertion rules out slippage-type natural insertion mechanisms
Because the PRRA insertion is out of frame relative to the aligned RaTG13/MP789 sequences, and insertions produced by polymerase slippage or release/repriming are expected to preserve the reading frame, those natural insertion mechanisms are excluded - leaving recombination or deliberate insertion.
Reasoning
Insertions generated by intramolecular polymerase slippage or by copy-choice release-and-repriming have been postulated to keep the viral reading frame intact, because they copy contiguous template in-frame. The SARS-CoV-2 PRRA insert, when aligned to RaTG13 and MP789, is out of frame - it splits an existing serine codon. Therefore the two commonly invoked spontaneous single-strand insertion mechanisms do not straightforwardly account for it. The remaining natural route is recombination/template switching with another virus supplying the insert; the alternative is deliberate insertion by site-directed mutagenesis - precisely the sort of edit routinely used to add furin sites to CoV spikes. This keeps a lab insertion a live possibility for the FCS.
Validity verdict — approved (checked)
Reconstruction. Premises: (i) polymerase-slippage and release/repriming insertions are expected to preserve the reading frame (copy contiguous template in-frame, respecting codon boundaries); (ii) the PRRA insert, aligned to RaTG13/MP789, is out of frame — it splits an existing serine codon. Load-bearing step: modus tollens — if mechanism M preserves frame and the observed insert does not preserve frame, M did not produce it — leaving recombination/template-switching or deliberate insertion.
Evaluation. The inference is a clean modus tollens and holds conditional on premise (i). The disjunctive conclusion (“recombination or deliberate insertion”) is not overclaimed. The obvious pressure point is premise (i) itself — whether slippage-type mechanisms really must respect codon boundaries (a 12-nt, i.e. in-frame-length, insert placed off codon phase is exactly what critics say slippage can do) — but that is premise truth, priced at step 8, not a failure of the inferential step. Given the premise as stated, no undercutting defeater survives. Approved; the logical step is elementary and traced → checked.
Link to original
A-70 - Rare CGG-CGG codon usage plus a conveniently placed FauI site is more expected from lab insertion
The FCS’s two arginines use CGG-CGG - a codon pair rare in these viruses and the only doubled CGG in the genome - and the insert conveniently carries a FauI restriction site usable for cloning/screening (RFLP), a pattern the authors argue is more readily explained by human codon choice and restriction-site design than by natural evolution.
Reasoning
In these coronaviruses arginine is only ~5% encoded by CGG, so an adjacent CGG-CGG pair - the sole doubled CGG in the entire genome - inside a brand-new insert is unusual for natural sequence, whereas researchers cloning from synthetic DNA commonly use whatever codons are convenient irrespective of host preference. In addition, the insert introduces a FauI restriction site; because the new furin site is prone to deletion in vitro, a nearby restriction site is just what would let a researcher use restriction-fragment-length-polymorphism methods to clone or screen for the insert. The authors argue the coincidence of an atypical codon choice and a functionally useful restriction site at the engineered locus fits a laboratory insertion better than spontaneous natural mutation.
Validity verdict — approved (checked)
Reconstruction. Premises: (i) arginine is only ~5% CGG-encoded in these CoVs, and the insert’s CGG-CGG is the sole doubled CGG in the genome; (ii) the insert introduces a FauI restriction site; (iii) the new furin site is deletion-prone in vitro, so a nearby restriction site is useful for RFLP-based cloning/screening. Load-bearing step: from “atypical codon usage + a functionally convenient restriction site at the insert” to “the pattern is more expected under deliberate lab insertion than natural evolution.”
Evaluation. This is a comparative-likelihood claim already pitched at the modest “more readily explained / more expected” level (not a proof), so there is no over-reach to correct down. Conditional on the premises the direction holds: designed sequences do use convenient (host-non-preferred) codons and researchers do exploit restriction sites, so P(features | lab) ≥ P(features | nature). Probed undercutting defeaters: (a) a doubled rare codon can arise naturally — 5% usage makes it unusual but far from impossible; (b) FauI (and some) restriction sites occur by chance every few hundred bp, so a site “near” the insert is close to guaranteed and thus nearly non-discriminating. These defeaters attack magnitude, and (b) drains most of the evidential weight from the FauI half, but neither reverses the direction; the codon-usage half still tilts toward design. So the argument is valid-but-weak, which is
Link to originalapproved(its strength — likely small — is priced at step 8), notcorrected, since the statement does not overclaim. Traced here → checked.
A-71 - Chimeric-CoV construction and furin-site insertion were routine, trace-free 2015-2018 techniques
Combining bat backbones with heterologous RBDs, inserting furin cleavage sites, and serial-passage adaptation in hACE2 animals were all established, published pre-2019 techniques (WIV/Baric gain-of-function work, the WIV1 eight-chimera panel, published FCS insertion, no-see-um seamless cloning) that leave no trace, so constructing SARS-CoV-2’s genome was technically feasible before the outbreak.
Reasoning
The paper documents two decades of chimeric-coronavirus construction: a 1999 feline/murine RBD swap; 2007 WIV ‘bat-man’ spike chimeras; a 2008 Baric live bat-backbone/SARS-RBD chimera; the 2015 SHC014 and 2016 WIV1 mouse-adapted-SARS-backbone chimeras; and a 2017 WIV panel of eight WIV1-backbone chimeras with transplanted bat RBDs. It also cites published methods for inserting polybasic furin cleavage sites into CoV spikes (increasing pathogenicity) and for serial passage in humanized/hACE2 animals. Because site-directed mutagenesis and seamless (no-see-um) cloning leave no residual scar, an engineered virus can be indistinguishable from a natural one. The upshot: the technical capability to assemble a RaTG13-backbone/pangolin-RBD chimera and to add a furin site existed and was in active use before 2019, so lab construction and serial-passage adaptation are feasible, not exotic, hypotheses. This argument bears on the prior plausibility of the lab-construction and serial-passage hypotheses.
Validity (step 6)
Reconstruction. Premises: each cited technique (heterologous-RBD backbone chimeras, polybasic FCS insertion, serial passage in hACE2 animals, no-see-um seamless cloning) was published/in active use pre-2019 and the seamless-cloning route leaves no residual scar. Load-bearing step: from “each component technique existed and is trace-free” to “assembling a chimeric SARS-CoV-2-like genome (and adding a furin site) was technically feasible before the outbreak.”
Verdict: approved (checked). Conditional on the premises, the feasibility conclusion follows — feasibility of the whole requires only that the component operations be available, which the premises grant; and the trace-free sub-premise licenses the separate “indistinguishable from natural” point riding on O-74. Probed for an undercutting defeater: one might object that possessing the operations does not establish possession of the specific RaTG13-like backbone + pangolin-like RBD starting material, so “feasible” could be inflated to “readily done.” But the statement’s conclusion is only technical feasibility / capability-in-principle, not that it was done, so that objection bears on truth/likelihood (priced at step 8), not on the inference. No defeater breaks the step at the stated modest strength.
Link to original
A-72 - Rebuttal- the RBD's human-ACE2 optimization does not preclude rational design
Andersen’s claim that a designer could not have predicted SARS-CoV-2’s high-affinity human-ACE2 RBD is weakened because structural analysis shows it recognises hACE2 more efficiently than SARS-CoV, chimera work shows bat spikes bind hACE2 with more plasticity than predicted, and WIV1/Rs3367 (which binds hACE2) already shares four of six critical RBD residues - so effective hACE2 binding was foreseeable/selectable in the lab.
Reasoning
Andersen et al. argued that because models built on the SARS-CoV RBD would not have predicted SARS-CoV-2’s binding solution, the RBD was unlikely to be designed. Segreto and Deigin counter that (1) Wan et al.’s structural analysis indicates SARS-CoV-2 recognises human ACE2 more efficiently than SARS-CoV; (2) gain-of-function chimera experiments demonstrated bat CoV spikes bind hACE2 with more plasticity than earlier predicted; and (3) the bat strain Rs3367/WIV1 - shown to bind hACE2 directly - already shares four of the six critical RBD residues with SARS-CoV-2, so a researcher could reasonably have expected a related RBD to bind hACE2. Consequently the ‘no one could have designed this’ premise does not hold, and rational design or lab selection of a high-affinity hACE2-binding RBD remains possible. This argument attaches to the lab-origin hypothesis by removing an objection to it, rather than by resting on one of this paper’s own observations.
Validity (step 6)
Reconstruction. This is a rebuttal (undercutter of Andersen’s “no designer could have predicted this RBD”). Premises: (1) SARS-CoV-2 recognises hACE2 more efficiently than SARS-CoV (Wan et al.); (2) chimera work shows bat-CoV spikes bind hACE2 with more plasticity than earlier predicted; (3) Rs3367/WIV1 binds hACE2 and already shares 4 of 6 critical RBD residues with SARS-CoV-2. Load-bearing step: from these, effective hACE2 binding was “foreseeable/selectable in the lab,” so Andersen’s premise is weakened (not that the RBD was designed).
Verdict: approved (checked). Conditional on the premises, the disjunctive conclusion holds. Probed defeater: Andersen’s actual claim was narrower — that computational modelling of the SARS-CoV-2 RBD would not have predicted its specific (non-theoretically-ideal) binding solution — so “more efficient than SARS-CoV” and “related strains bind hACE2” do not fully rebut the rational-design-from-models branch. But the conclusion is a disjunction that also contains “selectable in the lab,” and the lab-selection branch (serial passage / directed evolution) requires no ex-ante prediction at all, so it survives that defeater; premise (2)‘s plasticity plus a related hACE2-binder make selection of a high-affinity RBD a coherent route. Since the statement’s conclusion is only “weakened / foreseeable-or-selectable” — appropriately modest, not “refuted” — the step goes through. Truth of the biological premises is priced at step 8.
Link to original