iandunham / forge Goto Github PK

For details of the analysis approach see documentation in the web version at

http://www.1000genomes.org/forge-analysis

1. The script itself currently called forge.pl written in perl. It has
the following perl dependencies.

use 5.012;
use warnings;
use DBI;
use Sort::Naturally;
use Cwd;
use Storable;
use Getopt::Long;

2. The sqlite3 db file that stores the bitstrings. This is a 98Gb file for forge2.0
Called forge.db.

3. Four stored hashes containing the parameters for the background selection.
two files each for either the omni genotyping array, or all the other GWAS
snp arrays.
76M omni.snp_bins.10
231B omni.snp_params.10
63M snp_bins.10
231B snp_params.10

Both the database and the hashes are downloadable from


ftp://ftp.1000genomes.ebi.ac.uk/vol1/ftp/technical/browser/forge_11/
forge_db_20141112.tar.gz for forge1.1

ftp://ftp.1000genomes.ebi.ac.uk/vol1/ftp/technical/browser/forge/
forge_20131009.tar.gz for forge1.0

4. a forge.ini file in the same drectory as the script. Edit this to provide the
directory in which the dartabase and hashes are stored.

5. An R 3.0 installation with the "devtools" and "rCharts" packages installed. See

https://github.com/ramnathv/rCharts. You will need to install the latest version e.g.

require(devtools)
install_github('rCharts', 'ramnathv', ref = "dev")

The input data is one of several options.

a. A list of rsids for SNPs
b. a pseudobed format of chr\tbeg\tend\rsid
c. bed format of locations
d. vcf2.

The analysis requires a minimum of 20 SNPs (this is not a strict limit but
operationally is best).

To work SNPs have to be in phase 1 of 1000 genomes. The script gives warnings
on SNPs not found.

It also warns for background sets that do not have the right number of SNPs
chosen, but this is really for information only.

It takes a series of command line options as follows

-f : the file to run on
-data : whether to analyse ENCODE (encode) or EpigenomeRoadmap (erc) data
-label : a name for the files that are generated and for the plot titles
       where there is a title.
- format : for the input data formats. If this is location data e.g.
         bed or tabix or some vcf lines, the rsid is obtained from
         the sqlite3 database.
-bkgd : whether the background selection should be from Omniarray SNPs
         (omni) or the a general set of GWAS typing arrays (gwas).

Some of these default as described in the perldoc.6. Minimally the command
line is

forge.pl -f rsidfile -label Some_label

which will by default run on Epigenome Roadmap data with the gwas
background

OUTPUT
======
there are 3 outputs generated

1. A pdf static chart, that would be good for download.
2. Two alternate d3 interactive charts.  the *dchart.html are the best
in terms of axis labelling but have some minor quirks
3. A Datatables table.
4. There's also a tsv file of the results.