Exome-wide rare variant genetic load is associated with cognition, intellectual disability and psychiatric disorders in population-based cohorts
Verdict
UNSCIENTIFIC
Neither the paradigm nor the study inside it passes. The verdict is unscientific.
1·Study methodology · inside the paradigm
UNSCIENTIFIC
Does not follow the scientific method within Genomics / DNA Model.
2·Paradigm · Genomics / DNA Model
UNSCIENTIFIC
Fails the tests. The worse layer decides.
How the verdict is decided
The verdict grades both layers: the paradigm this study assumes, and how the study was carried out inside it. Both must be scientific for the whole thing to be scientific. A clean method inside an unscientific paradigm is unscientific. An unscientific method inside a better paradigm is unscientific. A split on either layer is mixed. The worse layer decides.
1·Study methodology
Did this study test its claim with methods that are independent, falsifiable, and non-circular?
UNSCIENTIFIC
This study claims that people carrying more rare broken versions of important genes score lower on thinking tests and are more likely to have autism, schizophrenia and intellectual disability. All three tests fail.
Independently Verifiable
Fail
Sequencers gave off light, software turned it into letters and a variant list, and the model's gene weights turned that list into a 'load' that exists only in the spreadsheet.
Falsifiable
Fail
A flat bipolar estimate, imprecise sibling results and a shrunken African-ancestry result were each explained away instead of counted against the claim.
Non-Circular
Fail
The damage scores, gene weights and 'established gene' lists all came from earlier disease-case studies, then were used here to score and split the same disorders.
Why
They took blood and spit from about 300,000 adults and 3,500 children and ran it through chemistry and sequencers that the model assumes are handling DNA. The machines gave off light, and software turned the light into letters and a list of letter changes. Other software built to rank variants by disease risk flagged some changes as damaging, and each gene hit got a number from earlier work based on how rare such changes are in other databases. They added those numbers into a 'load,' then set it against clinic billing codes, parent reports and one thinking-test score. Weak results were called imprecise. When African-ancestry results shrank, they switched to stricter diagnosis definitions and the numbers lined up. The scores and diagnoses are real. The load is the model adding up its own readings.
2·Paradigm · Genomics / DNA Model
Does the framework this study assumes pass the three tests?
UNSCIENTIFIC
Independently Verifiable
Fail
Sequencers produce light and software prints letters; a gene, a break in it, and a fitness cost for losing it exist only in the model's reading of those letters.
Falsifiable
Fail
Carriers who never get sick are called incompletely penetrant, and 99.5% of carriers here had no intellectual disability diagnosis, so unaffected carriers never count against the claim.
Non-Circular
Fail
The reference genome, the gene catalogue and the idea that a gene carries a code were assembled by the same field that now uses them to find disease genes.
Why
A fever breaking or a height on a tape can be witnessed. A gene cannot. The chain here ends in light on a detector that software prints as letters, so nothing a stranger could see closes it. Nobody has seen a gene, let alone one that carries a fitness cost. When carriers stay healthy, the field says incomplete penetrance, or a continuum from large to modest risk, as this paper's own introduction does. Any effect size and any non-effect fits that. The reference panels, gene lists and damage predictors came out of the same case-hunting work they now confirm. Predicting more intellectual disability among carriers does not make the gene claim observable.
From the study
For each participant, PTV and missense GLs were calculated separately as the sum of sHet values, as estimated by Sun et al.13, across all genes carrying at least one qualifying variant of that class, so that each affected gene contributed its selection coefficient once regardless of the number of qualifying variants it carried.
Because EHR-based schizophrenia and bipolar diagnoses may thus be recorded with different accuracy across groups, we additionally used more specific definitions in all ancestry groups.
In AoU, within-sibship estimates from conditional logistic regression in 4,229 full-sibling pairs from 3,883 families (Supplementary Note 3) were imprecise but did not differ significantly from the population estimates for any outcome (all P ≥ 0.16).
Assuming that these associations are causal, carrying a constrained PTV accounted for 21% of ID, 9% of schizophrenia and 6% of ASD cases in AoU, and for 1–5% of cases of the other outcomes.