This code enables users to get PICS credible sets for nearly any index SNP with a pvalue. Users can then test if two genetic signals colocalize in a Bayesian framework.
library("piccolo")
gwas_credSet <- pics.download(rsid="rs123", pvalue=-log10(1.23E-45))
eQTL_credSet <- pics.read("/full/path/to/pics.txt")
myColoc <- pics.coloc(gwas_credSet, eQTL_credSet)
See the pinned issue for a pptx showing further illustrating an example. Also available here
Test for colocalization of two PICS credible sets
| Options | Value | Default |
|---|---|---|
| data1, data2 | PICS sets from pics.read or pics.download | NA |
| pics1, pics2 | column name to pull PICS prob from. | "PICS_probability" |
| rsid1, rsid2 | column name to pull rsid from. | "Linked_SNP" |
| rounded | [#] Decimal points to round posteriors to. | |
| priorc1 | Prior probability for a siganl in data1 | 1e-4 |
| priorc2 | Prior probability for a siganl in data2 | 1e-4 |
| priorc12 | Prior probability for colocalization of both signals. | 1e-5 |
- Returns full coloc object.
- Users only need to specify the column names through pics[1|2] & rsid[1|2] if they are using credible sets that are not from pics.download().
- For "lighter" dataframe returing only H3 & H4 posteriors, use pics.coloc.lite()
- Adapted from Toby Johnson's code, which was adapted from Giambartolomei et al. 2014 colocalisation method
Test for colocalization of two PICS credible sets
- Returns a "lighter" dataframe with only H3 & H4 posteriors
Query the PICs website and get a PICs credible set for a SNP of interest
| Options | Value | Default |
|---|---|---|
| rsid | SNP rsID | Example: "rs1234" |
| pvalue | pvalue or -log10(pvalue). | Examples: -log10(1.23E-45) or 45 |
| ancestry | [EUR, ASN, AFR] | "EUR" |
| output | Full path and name for the download to be saved. | held in memory |
- Returns a dataframe with default colnames as pic[1|2]="PICS_probability" & rsid[1|2]="Linked_SNP" that pics.coloc anticipates as default.
- For pics.download() to work behind a proxy use this code in an R session in your ~/.Renviron.
options(RCurlOptions = list(proxy="proxy.URL:PORT", proxyuserpwd="ID:PW"))
Load a PICs file that was either download from the website (copy+paste to a text file) or a saved output from pics.download(output=X)
- If the rsID is not in the imputation used by PICs (1KGp1) then it returns the input rsID as the 1 causal SNP. This is almost certainly an erroneous credible set. This will also occur with rsIDs that do not exist. By default, pics.download() return NA.