liangyy / finemapaim

run-aim

Run AIM here.

Here we wrap it up as an R function taking genotype, ASC data, eQTL results and returning the PIP and CS. I mostly follow the paper "Methods" section "Application to GTEx data" but instead of using CAVIAR as fine-mapping solver we deploy susieR since it is much faster when we include all SNPs in the cis-window (thousands of SNPs). I use both of the meta-analysis scheme S^M1 and S^M2 as suggested by the paper. And we report the results of the scheme with smaller CS size on average.

Note

It relies on the current screen shot of susieR (610b5bb754e9c9f970eef15a5a194fb9dc421caa). It matters since it relies on susieR::susie_rss.