Colocalisation of GWAS/eQTL signals to identify shared biological mechanisms

 



This application takes p-values resulting from an association with a trait (a biomarker for example) and compares them with liver expression data outputting a posterior probability for a common signal within a region defined by the expression probe (200Kb upstream and downstream from the probe). To compare two input datasets against each other please use the R package coloc.

Things to know before running analysis:
  • Make sure to check the first seven options and change the values according to your input file. The last three input boxes include advanced options and should be changed only in specific cases.
  • Time it takes: it takes 1 minute or less to analyse imputed data from chromosome 1 and about 7 minutes to analyse imputed genome-wide data. If it takes longer than 15 minutes there will be an error: please go to the e-mail version of this website.
  • The input file must have markers according to the GRC37/hg19 human genome assembly.
  • The dataset is assumed to be a Caucasian population sample.
  • It is assumed that the causal variant is in high LD with SNPs in the dataset, so the method works best using imputed dataset. A number of tools including MINIMAC, BIMBAM, IMPUTE and PLINK are available for imputation.
  • The number 'PP.H4.abf' is the probability of a shared variant association.
Click here for a sample input file, containing p-values from TM Teslovich et.al and downloaded from this link
Click here to view the paper describing the method

Claudia Giambartolomei, Vincent Plagnol
UCL Genetics Institute
Website created using RWUI application.
If you have any questions or suggestions on how to improve this website, please send an e-mail to claudia.giambartolomei.10@ucl.ac.uk
Choose one expression dataset


Upload a .txt file containing markers and p-values

Tick if input file does not contain a header row

Sample size of dataset used to generate p-values

If the data used to generate p-values is a case-control sample, enter here the number of individuals that are cases. If not a case-control, leave as it is.

Column of dataset with names of markers

Column of dataset with p-values

Minimum posterior probability for common signal to be returned in results.     Warning: Set it to 0 to return all results (note: setting this to 0 takes longer and outputs a long list of results).




Advanced Options

Prior probability a SNP is associated with expression trait.     Warning: Leave set to default in most cases.

Prior probability a SNP is associated with input trait.     Warning: Leave set to default in most cases.

Prior probability a SNP is associated with both traits.     Warning: Leave set to default in most cases.