Those errors did not merely corrupt individual genome records.
According to the authors, “waves of systematic errors” followed waves of COVID variants and had “significant impacts on the inferred phylogenetic tree,” which is the computer reconstruction used to describe how SARS-CoV-2 allegedly evolved and how different samples were genetically related.
The study identifies how this happened: some software simply filled places where sequencing provided no data with letters copied from the SARS-CoV-2 reference sequence.
Because that reference is already the presumed ancestor, “we don’t know what was here” could become “the ancestral sequence was here.”
Those illegitimately computer-supplied letters could then become false evidence that SARS-CoV-2 had mutated back toward its presumed ancestor.
The researchers ultimately went back to the archived sequencing data behind millions of pandemic genomes and rebuilt them.
Afterward, large numbers of supposed evolutionary events disappeared, nearly 30,000 samples changed lineage classification, and one downstream analysis produced 1,400 fewer supposed separate introductions of SARS-CoV-2 into the United States.
The purported biological past had not changed.
The computer-generated genetic record used to reconstruct that past had.
That exposes a problem more fundamental than bad software.
If changing how computers constructed SARS-CoV-2 genomes could change mutations, lineage assignments, and reconstructed transmission history, those things were not simply observations of what the virus did.
They were conclusions drawn from computational reconstructions of sequencing data.
And the reconstructions themselves could be wrong.
The study therefore raises a question that applies far beyond the particular errors Hunt et al. corrected: “Where does the biological evidence end and the computer-generated story of SARS-CoV-2 begin?”
‘The Majority of SARS-CoV-2 Genomes’
Hunt et al. put the scale of the problem in unusually stark terms in the paper’s abstract:
“The majority of SARS-CoV-2 genomes obtained during the pandemic were derived by amplifying overlapping windows of the genome (‘tiled amplicons’), reconstructing their sequences and fitting them together. This leads to systematic errors in genomes unless the software is both aware of the amplicon scheme and of the error modes of amplicon sequencing. Additionally, over time, amplicon schemes need to be updated as new mutations in the virus interfere with the primer binding sites at the end of amplicons. Thus, waves of variants swept the world during the pandemic and were followed by waves of systematic errors in the genomes, which had significant impacts on the inferred phylogenetic tree.”
The passage establishes four facts at once.
- The majority of pandemic SARS-CoV-2 genomes were reconstructed from amplified, overlapping pieces.
- The process can lead to “systematic errors in genomes.”
- The authors say “waves of systematic errors” followed waves of variants.
- And those errors had “significant impacts” on the purported evolutionary tree.
That matters because the tree is not merely an illustration.
Genetic differences among reconstructed genomes are used to infer (assume) which samples are related, where mutations occurred, which samples belong to which lineages, and how those lineages allegedly descended from earlier ones.
So what?
Systematic errors existed in the evidence before that evidence was converted into evolutionary history.
The fundamental question is therefore unavoidable:
If the majority of SARS-CoV-2 genomes came through a reconstruction process capable of systematic error, and those errors significantly changed the inferred tree, how much of the purported COVID evolutionary history describes actual biology, and how much describes what the reconstruction system software falsely produced?
The Sequencing Data Provided No Answer
Hunt et al. went beneath finished GenBank genomes to examine the archived sequencing reads used to construct them.
In their supplementary analysis, Hunt et al. specifically searched for stretches of GenBank genomes where the archived sequencing reads provided zero coverage.
Meaning there were no sequencing reads showing what genetic letters were actually present there.
Yet the finished GenBank genome claimed definite letters matching the reference sequence.
The researchers said there was “no evidence from the reads to support a call of A, C, G, or T.”
In other words, the underlying sequencing evidence said:
- We don’t know what genetic letters were here.
Yet the finished genome said:
- These specific genetic letters were here.
Whatever confidence one places in genomic sequencing generally, these particular gaps present a simpler problem: “The sequencing data had provided no answer, but the finished genome contained one anyway.”
So what?
If no sequencing reads established whether a position contained A, C, G or T, that position could not establish whether one purported virus matched or differed from another there.
Yet genetic matches and differences are precisely what downstream evolutionary analyses use to reconstruct relationships and changes.
The Computer Supplied the Presumed Ancestor
If the underlying data did not establish what was there, how could a definite genetic sequence appear in its place and then legitimately become evidence about what SARS-CoV-2 supposedly did?
Hunt et al. identify how that happened.
Some genome-building software forced segments of the SARS-CoV-2 reference sequence to fill missing data.
The authors call the premise a “false assumption”: when a region contained no reads, software could nevertheless guess that the missing sequence matched the reference.
And the reference represented the presumed ancestral SARS-CoV-2 sequence in the evolutionary analysis.
The operation was therefore:
No sequencing evidence
? software inserts presumed ancestral sequence
? finished genome contains definite ancestral letters
? downstream analysis receives those letters as part of the genome.
The sample did not establish those letters at those positions.
The computer supplied them precisely because there was no evidence.
Hunt et al. say portions of the reference sequence had been used to fill dropouts for “a large number of sequences” and that this effect alone was a “significant cause of reversions in the tree.”
So what?
Researchers were attempting to infer how samples differed from their presumed ancestor while some genome-building software simply inserted that presumed ancestor into the sample wherever evidence was absent.
That reverses the normal direction of evidence.
Instead of the sample independently and directly establishing that it matched the presumed ancestor, the presumed ancestor itself supplied the match.
How can that computer-created similarity then legitimately count as evidence about ancestry or evolution?
Missing Evidence Became False Evolution
The consequences propagated into the evolutionary tree.
The mechanism is simple.
Presumed ancestor:
A
Purported descendant:
G
Later sample with no sequencing evidence:
?
Software supplies reference:
A
The completed genomes now appear to show:
A ? G ? A
The evolutionary analysis can infer that SARS-CoV-2 mutated away from its presumed ancestor and subsequently mutated back.
But the sample never established the final A.
The computer put it there.
Hunt et al. identify large numbers of reversions as an important signal of artifacts in pandemic evolutionary trees.
So what?
A purported event in SARS-CoV-2’s biological history could originate with information that was never obtained from the relevant portion of the sample.
The epistemic chain had become:
missing evidence ? computer-generated sequence ? inferred mutation ? purported viral history.
If an alleged evolutionary event exists because software first supplied the genetic state requiring that event, what evidence independently establishes that the purported virus itself ever underwent the alleged evolution?
Waves of Variants Produced Waves of Systematic Errors
Hunt et al. describe a problem that evolved alongside the purported virus.
The tiled-amplicon method is said to depend on primers working at particular genetic locations.
According to the authors, new mutations could interfere with those primer-binding sites, requiring the amplification schemes to be updated.
Hence their remarkable statement:
“waves of variants” were followed by “waves of systematic errors in the genomes.”
The error profile therefore was not necessarily constant throughout the pandemic.
It could change as the purported virus changed.
So what?
The same genetic changes researchers were trying to document could alter the performance of the system being used to document them.
That creates an epistemic feedback problem:
Researchers used the sequencing system to identify changing variants, while changing variants could themselves cause changing systematic errors in the sequencing-derived genomes.
When a new genetic pattern appeared during a variant wave, how confidently could downstream researchers distinguish a biological change in SARS-CoV-2 from a new systematic error produced because the purported biological change interfered with the system used to reconstruct its genome?
The Errors Entered the Evolutionary Tree
Hunt et al. make the downstream consequence explicit in their discussion.
They write that: “the downstream effect of using multiple variable-quality genome assembly workflows, inconsistent QC criteria and the inevitable coevolution of virus and amplicon schemas, led to systematic errors in genomes, and therefore the phylogeny.”
That sentence identifies where the errors went.
Into the genomes.
And therefore:
Into the evolutionary tree.
So what?
Once an erroneous genetic letter enters the finished genome, downstream analysis does not automatically know that the letter represents an illegitimate artifact rather than biology.
It can become another data point from which ancestry and evolution are inferred, not observed; then operationalized.
The authors themselves describe the problem as trying to distinguish “artifact from biology.”
This is big:
If researchers looking at the evolutionary tree had to determine whether something appearing in it represented “artifact” or “biology,” then the tree itself cannot simply be equated with an observed biological history.
Rebuilding the Genomes Erased Supposed Evolution
Hunt and his colleagues went back to the publicly archived sequencing data and rebuilt millions of purported SARS-CoV-2 genomes through a standardized pipeline called Viridian.
Then they compared the resulting evolutionary tree with the GenBank-based tree.
The GenBank tree contained 63 genome positions with at least 200 supposed reversions.
The Viridian tree contained:
20.
The purported viruses had not traveled back through time and changed.
The computational representation changed.
And large numbers of purported evolutionary events disappeared.
So what?
That means those events cannot all be treated as independently established facts about what SARS-CoV-2 biologically did.
At least some depended upon how computers converted sequencing data into finished genomes.
If the underlying biological past is fixed but changing the reconstruction changes the purported evolutionary history, which one is being reported when researchers say “the virus evolved”?
Nearly 30,000 Samples Changed Lineage
The consequences extended to the names assigned to purported viral lineages.
Hunt et al. found 29,475 samples with different Pango coronavirus lineage assignments depending on whether the GenBank genome or Viridian reconstruction was used.
The samples had not changed.
Their computer-generated genetic representation had.
And when that representation changed, nearly 30,000 samples changed classification.
So what?
“This sample was lineage X” sounds like an intrinsic biological fact discovered in the sample.
For these samples, however, the answer depended upon which reconstruction of the sequencing data was submitted to the lineage-classification system.
If a sample’s lineage can change when its computationally reconstructed genome changes, where does biological identity end and algorithmic classification begin?
The Reconstructed History of COVID Entering America Changed
The Nature Methods researchers then tested whether rebuilding the genomes changed something beyond the evolutionary tree: their reconstruction of how many separate times SARS-CoV-2 supposedly entered different countries.
Using the original GenBank genomes, Hunt et al.’s analysis inferred:
- 15,026 separate SARS-CoV-2 introductions into the United States.
Using the genomes Hunt et al. rebuilt from the archived sequencing data:
- 13,626.
The researchers therefore obtained 1,400 fewer inferred U.S. introductions after changing which reconstruction of the sequencing data they analyzed.
Hunt et al. explain that genome errors could separate samples that should have clustered together, causing their geographic analysis to infer additional artificial “introductions.”
Importantly, the team does not attribute all 1,400 specifically to the reference-filling error.
Their rebuilt genomes corrected multiple kinds of errors.
But the epistemic consequence is fundamental.
Neither 15,026 nor 13,626 represents researchers directly observing SARS-CoV-2 enter the United States that many times.
Both are histories inferred from genetic relationships among computationally reconstructed genomes.
Change the genomes, and the inferred history changes by 1,400 supposed introductions.
So what?
Errors originating in the construction of digital genomes were consequential enough to change what the same researchers’ model said had happened in the physical history of the pandemic.
If rebuilding the genetic record can erase 1,400 inferred entries of SARS-CoV-2 into the United States, how many other claims about where the virus supposedly came from, where it went and how it spread depend not on directly observed events, but on the particular computational reconstruction used to infer them?
The Fundamental Problem Is Bigger Than COVID
Hunt et al. frame Viridian partly as a solution for future epidemics and pandemics.
They specifically hope their work will reduce time spent examining trees and trying to distinguish “artifact from biology.”
That language exposes the deeper problem.
The final genetic record can contain both.
And researchers may have to determine afterward which is which.
That matters beyond SARS-CoV-2 because modern genetics routinely moves through layers: biological material produces instrument measurements; software turns measurements into genetic calls; computational pipelines assemble those calls into sequences; algorithms compare those sequences; and models infer relationships, ancestry, and evolutionary history.
Hunt et al. do not establish that genomic sequencing generally is invalid.
They establish something more precise.
And epistemically consequential: “A finished genome is not automatically identical to what the underlying evidence established, and a biological story inferred from that genome inherits errors introduced during its construction.”
The distinction becomes critical whenever language collapses those layers.
- A reconstructed sequence becomes “the genome.”
- A difference between reconstructed genomes becomes “a mutation.”
- A computational relationship becomes “ancestry.”
- An inferred tree becomes “evolutionary history.”
At what point in that chain did an instrument-supported observation become a computational inference?
And at what point did the inference begin being reported as though it were the biological event itself?
Bottom Line
The Nature Methods study shows that systematic errors entered SARS-CoV-2 genomes and significantly affected the alleged evolutionary tree built from them.
In one documented failure, software filled regions with the presumed ancestral sequence even though the underlying sequencing reads provided no evidence establishing what was there.
Those computer-supplied letters could then become false evidence of viral evolution.
When Hunt et al. rebuilt millions of genomes from the archived sequencing data, supposed evolutionary events disappeared, nearly 30,000 samples changed lineage classification, and their reconstruction produced 1,400 fewer inferred U.S. introductions.
The purported biological past had not changed; the computational genetic record had, and the history inferred from it changed with it.
The fundamental problem is therefore not merely that some genomes contained errors.
It is that computationally reconstructed genomes were treated as evidence for mutations, lineages, ancestry, evolution, and transmission history even though the study demonstrates that the reconstruction itself could create false biological evidence.
That leaves the central question:
How much of COVID’s purported genetic backstory/alibi was established by the underlying biological evidence, and how much was produced by the computational systems used to turn that evidence into genomes, evolutionary trees and claims about what SARS-CoV-2 supposedly did?
Your support is crucial in helping us defeat mass censorship. Please consider donating via Locals or check out our unique merch. Follow us on X @ModernityNews.