Chipmatch

Chipmatch is a tool that identifies which chip type has been used for a given PLINK dataset. The analysis is based on Will Rayner’s Genotyping chips strand and build files.

As of 2018-11-06, his website contains 411 usable annotation files. Analyzing 650k variants against all 411 chip/build combinations usually takes less than 10 minutes.

Compiling

For compilation, you will need a working cargo/Rust installation. If you install rust through the Rustup Installer, you will already have a sufficiently recent Rust version. After that `cargo build –release` will get you a binary.

Chances are that Jan <j.kaessens@ikmb.uni-kiel.de> has a pre-compiled binary on the rzcluster.

Preparation

Download all the ZIP files from Will Rayner’s Website, or at least those you’re interested in, and dump them to a folder. Do not extract them.

Running

$ chipmatch --help
chipmatch 0.4.1
Jan Christian Kaessens <j.kaessens@ikmb.uni-kiel.de>
Tries to identify the chip type from a given BIM file and checks for orientation

USAGE:
    chipmatch [FLAGS] [OPTIONS] <FILE> <DIR>

FLAGS:
    -h, --help       Prints help information
    -V, --version    Prints version information
    -v, --verbose    Be verbose and print progress

OPTIONS:
    -e, --extract <N>      Extract the N most promising strand files to the local working directory
    -o, --output <FILE>    Write result table to FILE instead of stdout
    -t, --threads <N>      Use N threads for computation

ARGS:
    <FILE>    PLINK .bim file to guess the chip type for
    <DIR>     Directory containing Will Rayner's strand archives

The program’s output consists of three columns:

The annotation file name (“strand file”) as it appears in the ZIP archive
Name match, the ratio of SNPs where a corresponding SNP ID could be found
Name/Pos match, the ratio of SNPs with name vs. SNPs with matching name and position
Original strand rate, the ratio of SNPs with name and pos, that also have the same alleles (non-AT/CG)
Plus strand rate, the ratio of SNPs with name and pos, that also have the same alleles, flipping the reference to “+” (non-AT/CG)
AT/CG rate, the ratio of AT/CG SNPs in all SNPs with matching name and position

The table is sorted by match rate and in descending order.

Without the `–verbose` switch, only the tab-separated table is printed, with header line, making it easily parseable for script and pipeline embedding.

Example output:

$ chipmatch ../GSA-data/DKTNF.bim ../GSA
GSA-24v1-0_A6-b37.strand	0.9057662	0.9865884
GSA-24v1-0_A2-b37.strand	0.9057662	0.9865884
GSA-24v1-0_A1-b37.strand	0.9057662	0.9865884
GSA-24v2-0_A1-b37.strand	0.876778	0.92227954
GSA-24v1-0_A2-b38.strand	0.008157662	0.008886434
GSA-24v1-0_A1-b38.strand	0.008157662	0.008886434
GSA-24v1-0_A6-b38.strand	0.008157662	0.008886434
GSA-24v2-0_A1-b38.strand	0.0079119755	0.008328985
GSA-24v1-0_A2-b36.strand	0.00016569582	0.00018057342
GSA-24v1-0_A1-b36.strand	0.00016569582	0.00018057342
GSA-24v1-0_A6-b36.strand	0.00016569582	0.00018057342
GSA-24v2-0_A1-b36.strand	0.00015426852	0.00016235182

Verbose Example output