Skip to content

Latest commit

 

History

History
51 lines (31 loc) · 3.22 KB

README.md

File metadata and controls

51 lines (31 loc) · 3.22 KB
Raw

PAINTOR

Probabilistic Annotation INtegraTOR

##UPDATE 07/16/2015 Software (v2.1) has been updated with new strategy of setting Non-centrality Parameters (NCPs). See User Manual for details.

UPDATE 06/08/2015

Announcing PAINTOR v2.0! We have expanded PAINTOR functionality to conduct fine-mapping with multi-ethnic cohorts. This latest release includes updates to the input formats and command line flags. Please see the new User Manual for complete details.

Description

We provide a command line implementation of the PAINTOR and PAINTOR Trans-ethnic frameworks described in Kichaev et al. (PLOS Genetics, 2014) and Kichaev and Pasaniuc (American Journal of Human Genetics, 2015). Briefly, PAINTOR is a statistical fine-mapping method that integrates functional genomic data with association strength from potentially multiple populations to prioritize variants for follow-up analysis. The software runs on multiple fine-mapping loci and/or populations simultaneously and takes as input the following data for each set of SNPs at a locus

  1. Summary Association Statistics (Z-scores)
  2. Linkage Disequilibrium Matrix/Matrices (Pairwise Pearson correlations coefficients between each SNP)
  3. Functional Annotation Matrix (Binary indicator of annotation membership (i.e. if entry {i,k} = 1, then SNP i is a member of annotation K)

Key Features

  1. Outputs a probability for a SNP to be causal which can subsequently be used to prioritize variants
  2. Can model multiple causal variants at any risk locus
  3. Leverage functional genomic data as a prior probability to improve prioritization
  • This prior probability is not pre-specified, but rather, learned directly from the data via an Empirical Bayes approach
  1. Quantify enrichment of causal variants within functional classes
  • Enables users to unbiasedly select from a (potentially) large pool functional annotations that are most phenotypically relevant
  1. Model population-specific LD patterns.

Installation

The software has two dependencies: [1] Eigen v3.2 (matrix library) [2] NLopt v2.4.2 (optimization library) which are packaged with PAINTOR in order to simplify installation. Please see the Eigen homepage and NLopt homepage for more information.

Download the latest version of the software into your desired target directory. Then unpack and install the software with the following commands:

$ tar -xvf PAINTOR_FineMapping

$ cd PAINTOR_FineMapping

$ bash install.sh

This will create an executable "PAINTOR". Sample data is provided with the package. To test that the installation worked properly, type:

$ ./PAINTOR -input SampleData/input.files -in SampleData/ -out SampleData/ -Zhead Zscore.p1,Zscore.p2,Zscore.p3 -LDname LD.p1,LD.p2,LD.p3 -c 2 -annotations Coding,DHS

If everything worked correctly the final log-likelihood should be: -2115.89773

For quick start simply type:

$ ./PAINTOR

For detailed information on input files and command line flags see the user manual provided.