A gene-ranking tool that disagrees with GWAS, and 21 asthma genes to argue about

The swap here is narrow and worth stating precisely. Existing prioritization tools assume the causal gene sits close to the associated variant, and Xuanyao Liu's complaint is that when you search around the variant you often do not find much ^[9]. Her account puts the driver downstream in a trans-regulatory network, possibly on a different chromosome, on the receiving end of an effect originating far away ^[10]. The paper formalizes that as disease-proximal genes: the ones that centrally mediate the effects of other, more distal disease-associated genes ^[7]. Every claim the method makes rests on that single relocation of where causality is expected to live.

What makes it a decision tool rather than another score is that the ranking was put somewhere it could fail. CRISPR screens and mouse models were used to check that the prioritized genes produce asthma phenotypes ^[2], which is the step the field has lacked: the prevailing view already held that a small set of central genes matter most, with no reliable way to pick them out and test them experimentally ^[11].

The missing figure is the denominator. Twenty-one genes came out validated ^[1], and the account as supplied does not say how many candidates DANDELION ranked highly to begin with, or how many failed the screens ^[16]. That number is the whole economics of shortlisting. If "most" of the 21 were missed by other methods ^[1], then at least eleven genes on this list have no GWAS or PoPS support behind them ^[14], and two of the 21, under a tenth of the list, currently share a mechanistic story: a fatty acid metabolism and protein palmitoylation pathway that has not yet been studied for asthma ^[15][3].

That is the awkward part for anyone writing the internal case. Novelty against GWAS and PoPS is the selling point ^[4] and simultaneously the reason there is no second, independent line of evidence pointing at these genes. The authors argue GWAS is structurally unsuited to this job, being weighted toward common variants with small effect sizes that selection makes unlikely to touch critical pathways ^[8]. If that argument is right, the existing literature is a poor referee, and the mouse and CRISPR data have to carry more weight than validation data usually does. A target committee that demotes a variant-adjacent candidate on the strength of this will be leaning on one team's experiments in one disease.

The scope is honest enough in the paper's own title, which is about asthma rather than about disease in general ^[6]. The team, led from the University of Chicago and Columbia and publishing in Cell, has demonstrated the tool once, on one condition ^[5]. The name comes from the seed head of the flower, a branching network pointing inward to a center ^[12], which is a fair description of the assumption and also a reminder that the assumption has been tested on exactly one plant.