Skip to content
/ idul Public

iterative dispersion update to fit linear mixed model

License

Notifications You must be signed in to change notification settings

haplotype/idul

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

58 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

IDUL

IDUL (iterative dispersion update to fit linear mixed model) is designed for multi-omics analysis where each SNPs are tested for association with many phenotypes. IDUL has both theoretical and practical advantages over the Newton-Raphson method.

Current version

Version 0.83 was first compiled on 23 May 2024. Linux exectuable can be downloaded from above, or from http://www.haplotype.org/download/idul.linux.0.83.tar.gz. An folder contain example files can also be downloaded from above. You may also choose to compile from the source code in src/, but you want to install eigen and boost packages first.

A preprint documenting IDUL can be found here: https://biorxiv.org/cgi/content/short/2023.10.25.563975v3. Another preprint about computing Bayes factors for linear mixed model can be found here: https://biorxiv.org/cgi/content/short/2024.05.28.596229v1.

Usage exmpales

  1. This example takes vcf file as genotypes input. The output contains only likelihood ratio test p-values. ./idul -i input.vcf.gz -p phenotypes.gz -k kinship.txt -o pref

  2. This example takes bimbam mean genotype as genotype input. Since -b is invoked, it computes and outputs (log) Bayes factors.
    ./idul -b -g input.bimbam.mgt.gz -p phenotypes.gz -k kinship.gz -o pref

  3. This example computes residual. It takes covariates file, phenotype file and and kinship file as input, and output residuals.
    ./idul -r -c cov.txt -p phenotypes.gz -k kinship.gz -o pref

Input and options

y = W a + x b + u + e 

This is the standard linear mixed model. The following options specify each components (except e).

-p phenotype_file (y). 
-c covariate_file (W). 
-i vcf_file  (x).  
-g bimbam_mean_genotypes (x). 
-k kinship_file (u).  

Phenotypes file is expected to have multiple columns (a single column is okay too) with one column for each phenotype. Phenotypes should have no missing values. Covariates file is a rectangle, with the same number of rows as phenotypes, and however many columns needed. Missing values are allowed for vcf files (replaced with the mean genotypes), but not allowed for bimbam mean genotypes. Kinship is a square and symmetric kinsihp matrix, such as one estimated by kindred. Rows of bimbam mean genotype looks like below.

rs123 A G 0 1 2 2 0 1.1 0 
rs456 C T 1 0 1 0.05 0 2 0 

The options related to input files:

-e eigenvector_file.   
-v eigenvalues_file.  
-f minor_allele_freq. 

In eigenvector file, each column is an eigenvector. Eigenvalues file contains one row with n numbers, its jth number should correspond to j-th column of the eigenvectors. Two options -e and -v come together, and mutually exclusive with -k. SNPs whose minor allele frequencies are below the threshold will be removed.

Output and options

 -o output_pref. 
 -a output additional test statistics. 

There are three output files: pref.log (a txt document contains log), pref.snpinfo.txt.gz, and pref.pval.gz file, where pref is specified by -o. There are four columns in pref.snpinfo.txt.gz, SNP ID, A-allele, B-allele, and MAF (minor allele frequency). Since Ver 0.81, IDUL only output p-values by default. If -a was invoked, output contains additional test statistics such as beta, sigma, l0 (loglikelihood of null), l1 (loglikelihood of alternative), h (eta/(eta+1)), and niter (number of iterations used in optimization).

Other options

   -t  number_of_threads.  
   -w  compute wald test p-values instead of likelihood ratio test p-values. 
   -x  use IDUL+ algorithm instead of IDUL. 
   -b  compute Bayes factors. 
   -r  compute residual of phenotype files. 

About

iterative dispersion update to fit linear mixed model

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published