pnrobinson
commented
6 years ago The kind of strategy you are referring to is similar to the genic intolerance project: http://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1003709 Another related approach uses population constraint in the human population to assess categories of human functional elements (including noncoding). This is a nice paper https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4787796/ and is related to the Science paper after years prior from the Gerstein group that was followed up by this (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4203974/). There is an issue of statistical power since there are only a relatively small number of FA genes. I am a little sceptical that it will be possible to robustly find genes whose similarity to FA genes in this respect will robustly identify novel biology. However, it is probably worth eyeballing the data for a week or so and trying to figure out if there is anything to be done. I would recommend studying the work of the Goldstein and Gerstein groups cited above, they are both quite elegant and good places for us to start.
Complementary to looking for evolutionary variation in #35, we could query for degree of human variation position and frequency in the upstream regions of genes identified as being similar to our FA genes.
There may be some risk metric that could combine these two - essentially regions that show low variation in human and across evolution and then show a variant in our St. Judes data set might be more likely to be a shooting gun for other phenotypes/related diseases, essentially modifiers of FA.
Not sure how much frequency info we'd have but worth trying I think? and/or we can generate our own frequency from our cohort too.
@dnahotline @pnrobinson what do you think?