Embedded Files
MODE - Multivariate Optimization for Detection and Estimation is an open source C package developed for fine-mapping and effect size estimation of genetic causal variants. The method uses non-linear conjugate gradient method to simultaneously determine the probability of a SNP being causal and estimate its magnitude. The function minimized in the optimization routine is given below.
The parameters of the optimization (prior probability of a SNP being causal and its effect size) are modeled using a set of correlated non-Gaussian random variables. Using the Nataf's transformation, the random variables are transformed into standard normal space and the non-linear conjugate gradient method is used to efficiently obtain the estimates of probability of causality and effect sizes. The software is released under the GNU-GPL license.
Citation:
If you use MODE in any published work, please cite the software as an electronic resource/URL
V. S. Sundar, 2018, MODE for identifying genetic causal variants, available from https://sites.google.com/site/sundarvelkur/multivariate-optimization-for-detection-and-estimation.
Download:
Linux mode-1.00-x86_64
Input flags
-nsubj number of subjects
-nsnp number of SNPs
-gpmat Genotype-Phenotype matrix with nsubj rows and (1+nsnp)columns; the first column is the phenotype vector.
Genotype matrix should be coded as (0,1,2) with -9 representing missing values, or should be normalized.
-type type of Phenotype - binary, quantitative
-pp nsnp x 2 matrix of parameters a and b, assuming that the prior probabilities are distributed as U[a,b]; default U[0,1]
-es nsnp x 2 matrix of parameters mu and sigma, assuming that the effect sizes are distributed as N(mu,sigma); default N(0,1).
-ppcorr nsnp x nsnp probability correlation matrix
-escorr nsnp x nsnp effect size correlation matrix
-ppescorr nsnp x nsnp probability and effect size correation matrix
-h2 assumed heritability - (0,1] (default 1)
-sig0 shrinkage parameter - (0,1] (default 0.1)
standard deviation of the laplace probability density function
-nopt number of optimization iterations (default 1000)
-o output file to write probabilities and effect sizes (default results.txt)
-normalized 1 if the Genotype matrix is normalized
0 if the Genotype matrix is coded as (0,1,2), with -9 representing missing values
The first four flags, -nsubj, -nsnp, -gpmat,-type are mandatory.
Format
case-control study for 5 subjects and 6 SNPs
Genotype-Phenotype matrix (5 x 7 matrix).
The first column is the disease status 0 or 1.
1 2 0 0 0 1 1
0 0 1 1 0 0 0
0 1 1 0 0 0 1
1 0 0 0 1 1 1
1 1 1 2 1 0 1
Correlation among prior probabilities (6 x 6 matrix)
1 0.1 0 0 0 0
0.1 1 0 0 0 0
0 0 1 0 0 0
0 0 0 1 0 0
0 0 0 0 1 0
0 0 0 0 0 1
Correlation among effect sizes (6 x 6 matrix)
1 0 0 0 0 0
0 1 0 0 0 0
0 0 1 0 0 0
0 0 0 1 0 0
0 0 0 0 1 0.2
0 0 0 0 0.2 1
Correlation among prior probabilities and effect sizes (6 x 6 matrix)
1 0 0 0 0 0
0 1 0 0 0 0
0 0 1 0 0 0
0 0 0 1 0 0
0 0 0 0 1 0
0 0 0 0 0 1
Example (data)
mode -nsubj 1000 -nsnp 1000 -gpmat X_1000_1000.txt -type quantitative -h2 0.5 -nopt 5000 -normalized 1
Page updated
Google Sites
Report abuse