Plink

Home

This page says how to run plink to get eQTL results for Circos Plots

Table of Contents Step 1: Obtain ped and map files Step 2: Obtain other input files Step 3: Create SNP table in Oracle Step 4: Run perl script Step 5: Copy Data to other Databases

Step 1: Obtain ped and map files

Laura Saba supplies the ped and map files. These files should be placed in the directory

 /data/clemensl/plink/pedmapfiles

to provide some speedup you should convert to binary ped files(.bed not to be confused with ucsc .bed files)

 ./plink --file hxb.rn5 --map3 --make-bed

Step 2: Obtain other input files

Laura Saba supplies the input files, for example HXBBXH.brain.ensemblTrans.txt. These files should be placed in the directory

or

Generate the Group means by running method in output.Raw.Specific.Gene.Both.HDF5_mod.R Open the text output and remove Std Error columns and make sure Group Names match the group names in the ped file.

On a Mac this was done in Excel but be sure to save as windows formatted txt instead of just a txt.

This step requires an HDF5 file with all of the data required. Create using ConvertToHDF5 netbeans project.

Save files to:

 /plink/inputfiles

Step 3: Create SNP table in Oracle

In this step, use the text file(HXB.eQTL.masterList.rn5.txt) to populate a SNPS_NEW table in Oracle on Dev. Then copy the new SNPS to other DB by exporting table to inserts .sql.

Use EQTLLoader netbeans project to load DB.

Step 4: Run perl script

Run the perl script runBatch.pl which takes the line count of the input txt file and splits it evenly between x# processes to run plink repeatedly and places the resulting data into the Dev database in location_specific_eqtl_new. You can then copy from dev to the other databases. Alternately you can uncomment the output line and output to a file by editing runPlinkDB.pl.

runBatch.pl takes command line arguments for all required inputs.

run using:

 ./perl runBatch.pl brain hxb.brain.txt locusSpecPval hxb.bxh.rn5 Rat 4

0 - tissue

1 - Input Text File with probeset/transcript data

2 - Output File prefix(not used if loading into DB)

3 - Ped/Map file prefix

4 - Species

5 - Number of processes to start(I've run up to 24 without slowing things down too much on phenogen)

Step 5: Copy Data to other Databases

In SQLPLUS:

 SET COPYCOMMIT 5000;
 copy FROM user/password@sourceDB to user/password@destDB INSERT location_specific_eqtl_new (PROBE_ID,SNP_ID,PVALUE) USING (Select PROBE_ID,SNP_ID,PVALUE from location_specific_eqtl_new where snp_id>#lowest_SNP_ID);

setting copycommit is required or it will just do one commit at end and probably crash the DB.

Set the snp_id>9999 to a value just below the lowest snp_id added. It is important to move snps from Dev to others since the snp_ids for the data will remain constant since they are just copied from one DB to another.