PLINK file formats

Data in binary PLINK format, such as that provided for this tutorial, consists of a set of three files each containing information about the genotyping data and the samples included in the dataset. A brief description of each is provided below. More information on file formats can be found on the PLINK website

.bed files

The binary .bed file contains the genotype calls for each individual in the dataset. As a binary file, data is encoded as bytes, and not easily readable except through programs such as PLINK.

.bim files

These text files contain information about each variant in the dataset. The file consists of one line per variant, with the following columns:

1. Chromosome name
2. Variant identifier
3. Genomic position in morgans or centimorgans
4. Base-pair position (1-based)
5. Allele 1 (usually the minor allele)
6. Allele 2 (usually the major allele)

Note

PLINK currently does not handle multi-alellic SNPs (those with more than one alternate allele)

It also has a bad habit of 'flipping' data so the first allele is the minor allele (in the data). If you want to avoid that, you can run all the commands with the --keep-allele-order option, but we've skipped that in this tutorial.

.fam files

These contain information about each sample in the dataset. The file consists of one line per sample, with the following columns:

1. Family ID
2. Within-family ID (Individual ID)
3. Paternal ID (or "0" if not in dataset)
4. Maternal ID (or "0" if not in dataset)
5. Sex ("1" = male, "2" = female, "0" = unknown)
6. Phenotype value ("-9" = missing; in this data, phenotypes are provided in a separate file)

.qassoc files

These contain the summary statistics for each SNP included in the association analysis. The file has a header line followed by a set of statistics for each SNP. The column names and contents are as follows:

column	description
CHR	Chromosome code
SNP	Variant identifier
BP	Base-pair coordinate
NMISS	Number of nonmissing genotype calls
BETA	Regression coefficient
SE	Standard error
R2	Regression r-squared
T	Wald test (based on t-distribution)
P	Wald test asymptotic p-value