Skip to content

LD File Formats

Jump to bottom
rkwalters edited this page Feb 1, 2016 · 6 revisions

This page describes the file formats required for estimating and storing LD Scores.

There are three main file formats:

  1. .bed/.bim/.fam -- see the Plink 1.9 documentation
  2. .annot.
  3. 'l2.ldscore, l2.M_5_50

NOTE chromosomes are assumed to be integers. We haven't yet implemented LD Score regression for sex chromosomes

.annot

For annotations. One row per SNP. Whitespace-delimited text, with header. ldsc supports compression either with gzip or bzip2. Column order matters. The columns are

  1. CHR -- chromosome
  2. BP -- physical position (base pairs)
  3. SNP -- SNP identifier (rs number)
  4. CM -- genetic position (centimorgans)
  5. all additional columns -- Annotations

Annotations can be binary (e.g., DHS yes/no) or continuous (e.g., DAF, McVicker B).

Example:

CHR BP SNP CM AN1 AN2
1   1  rs1  0  1  0
1   2  rs2  0  1  0
1   3  rs3  0  0  1

.l2.ldscore

For LD Scores. This is the default output format of the --l2 flag. One row per SNP. Whitespace-delimited text, with header. ldsc supports compression either with gzip or bzip2, with gzip as the default. Column order matters. The columns are

  1. CHR -- chromosome
  2. BP -- physical position (base pairs)
  3. SNP -- SNP identifier (rs number)
  4. CM -- genetic position (centimorgans)
  5. MAF -- minor allele frequency
  6. all additional columns -- LD Scores

The CM column can be safely set to zero. Genetic positions are only used by the --ld-wind-cm flag, which reads cM coordinates from the .bim file.

Example:

CHR     SNP     BP      CM      MAF     LD
22      rs146752890     16050612        0       0.0672  71.06
22      rs139377059     16050678        0       0.0553  100.0
22      rs6518357       16051107        0       0.0593  108.6
22      rs62224609      16051249        0       0.0897  20.12
22      rs62224610      16051347        0       0.3337  19.59
22      rs143503259     16051453        0       0.0936  21.34

l2.M_5_50

One line, # of columns = number of annotations in the accompanying .l2.ldscore file in the same order. Each column contains the number of SNPs in the corresponding annotation category with MAF > 5%. The .l2.M file format is the same, except without the restriction on MAF.

Example:

100	1000
Clone this wiki locally