Loss-of-function * risk between EDDM3A and Alzheimer disease

EDDM3A was labelled as loss-of-function and risk factor for Alzheimer disease, but the model is from synonymous. I am wondering how such decision is made which sounds pretty weird.

See page here

Thanks @Shicheng_Guo for the feedback. We are eager to get more feedback in the direction of effect and this discussion is very useful to understand what to expect from different types of evidence. We are already working on some improvements and some assessments will change for our next release.

The case you are bringing up is an interesting one and we had a brief discussion about it in the past but it would be interesting to hear your thoughts. My understanding is that the occurrence of rare synonymous variants in EDDM3A qualifies as increased risk of suffering Alzheimer Disease. Independently of the gene effect, I understand that there should be a mechanism by which synonymous mutations in this gene can cause the phenotype. The only mechanism that comes to my mind is a shift in the codon usage that triggers problems in the translation of this protein. If this was the case, I think it would qualify as a loss of function and the assessment would be correct.

There is also a risk the evidence is just an artefact of the burden test and therefore not relevant at all, but I guess this is independent of the direction of effect assesment.

Does this sound reasonable? What would you expect for this particular example?

Intrestingly there’s a paper discussing just that: Codon bias among synonymous rare variants is associated with Alzheimer’s disease imaging biomarker

The study is based on whole-genome sequencing of AD patients. The collapsing test (SKAT-O) found:

While using all synonymous variants we did not identify any genome-wide significant associations, using only synonymous variants that affected codon frequency we identified several genes as significantly associated with the imaging phenotype. Additionally, significant associations were found using only rare variants that contains an optimal codon in among minor alleles and a non-optimal codon in the major allele.

To me it suggests that the risk only increases when the rare codon is replaced with the common codon. They are arguing that this replacement negatively impacts the folding, meaning a loss of function effect. However to me it is too speculative (they also don’t provide any experimental evidence). I can imagine a scenario, when a rare to common codon change might increase the efficiency of the translation, hence can be interpreted as gain of function event, where an antagonistic targeting strategy would be advantageous.