README.md

Match Results

the WMDA results look like this P000001;D000001;M;0;M;0;M;0;M;0;M;0;0;0 1000 patients x 10000 donors = 1M rows

The single-locus match grade values are A, M, P The single-locus 2-allele match likelihoods range from 0-100 A => always 100 M => always 0 P => ranges 0-100

Perl code

• generate_matchgraph.pl creates the graph in ./graph
• bulk_load_neo4j.sh loads the graph from ./graph
• match_results.pl computes the results in a format that can be compared to ../graph_generator/data/wmda/set3.consensus.txt

Walk-through of Running Match

Pre-requisites

• Generate Imputation files for both Patient/Donors. See (Multi-race Imputation)[../../multi_race_impute]

• Perl

  • Install Perl Brew from http://perlbrew.pl/

  curl -L https://install.perlbrew.pl | bash
  

  Add perlbrew to your bashrc

  source ~/perl5/perlbrew/etc/bashrc
  

  Install the latest stable build of Perl. This could take sometime.

  perlbrew init
  perlbrew install perl-5.26.0
  

  Use the installed Perl

  perlbrew list
  perlbrew use perl-5.26.0
  
  perl --version
  

  Install the Perl libraries used by matching.

  cpan install REST::Client
  cpan install JSON
  cpan install JSON::Parse
  cpan install Math::Round
  

Matching

Generate the CSV files for loading into Neo4J

cd matching/graph_generation/perl
mkdir -p output/graph
time ./generate_matchgraph.pl

Load the CSV files to Neo4J

time ./bulk_load_neo4j.sh

Visit http://localhost:7474/browser/ to verify database is working.

Run the matcher program from grimm/matching/search.

cd ../../search
mkdir -p output
time ./match_results.pl

This produces a output/mr.txt file that holds the match results.

ls -lah output/mr.txt

Compare the results from the WMDA consensus results.

time ./compare_results.pl

This will produce a output/cmp.txt file. This will list the difference in the Neo4J version versus Consensus rsults.

wc -l output/cmp.txt
       0 output/cmp.txt

This produces produce zero different comparisons for WMDA data.

0 output/cmp.txt