Revision as of 11:49, 25 June 2013

This will contain the program called NGSadmix, which is a very nice tool for finding admixture. It is based on genotype likelihoods or genotype probabilities. It is a fancy multithreaded c/c++ program

Installation

wget popgen.dk/software/NGSadmix/ngsadmix32.cpp 
g++ ngsadmix32.cpp -O3 -lpthread -lz -o NGSadmix

Input Files

The current input files are the widely used beagle inputfiles, or beagle imputed outputfiles [1]. We recommend ANGSD for easy transformation of Next-generation sequencing data to beagle format.

Options

./NGSadmix
Arguments:
	-likes Beagle likelihood filename
	-K Number of ancestral populations
Optional:
	-fname Ancestral population frequencies
	-qname Admixture proportions
	-outfiles Prefix for output files
	-printInfo print ID and mean maf for the SNPs that were analysed
Setup:
	-seed Seed for initial guess in EM
	-P Number of threads
	-method If 0 no acceleration of EM algorithm
	-misTol Tolerance for considering site as missing
Stop chriteria:
	-tolLike50 Loglikelihood difference in 50 iterations
	-tol Tolerance for convergence
	-dymBound Use dymamic boundaries (1: yes (default) 0: no)
	-maxiter Maximum number of EM iterations
Filtering
	-minMaf Minimum minor allele frequency
	-minLrt Minimum likelihood ratio value for maf>0
	-minInd Minumum number of informative individuals

Output Files

Program outputs 3 files.

PREFIX.log
PREFIX.fopt.gz
PREFIX.qopt

Log file

	-> Dumping file: tskSim/tsk6GL.beagle.s1.log
	-> Dumping file: tskSim/tsk6GL.beagle.s1.filter
Input: lname=tskSim/tsk6GL.beagle nPop=3, fname=(null) qname=(null) outfiles=tskSim/tsk6GL.beagle.s1
Setup: seed=1 nThreads=10 method=1
Convergence: maxIter=2000 tol=0.000000 tolLike50=0.010000 dymBound=0
Filters: misTol=0.050000 minMaf=0.000000 minLrt=0.000000 minInd=0
Input file has dim: nsites=100000 nind=75
Input file has dim (AFTER filtering): nsites=100000 nind=75
iter[start] like is=9299805.984931
iter[50] like is=-6531138.892608 thres=0.002800
iter[100] like is=-6528710.773349 thres=0.001289
iter[150] like is=-6528405.896951 thres=0.001211
iter[200] like is=-6528306.803820 thres=0.000420
iter[250] like is=-6528277.160993 thres=0.000546
iter[300] like is=-6528271.925055 thres=0.000033
iter[350] like is=-6528271.177692 thres=0.000008
iter[400] like is=-6528270.876315 thres=0.000005
iter[450] like is=-6528270.772894 thres=0.000140
iter[500] like is=-6528270.747721 thres=0.000002
iter[550] like is=-6528270.740654 thres=0.000002
Convergence achived because log likelihooditer difference for 50 iteraction is less than 0.010000
best like=-6528270.740654 after 550 iterations
	-> Dumping file: tskSim/tsk6GL.beagle.s1.qopt
	-> Dumping file: tskSim/tsk6GL.beagle.s1.fopt.gz
	[ALL done] cpu-time used =  671.82 sec
	[ALL done] walltime used =  114.00 sec

log

v32 june 25-2013; modified code such that it now compiles on OSX
v31 june 24-2013; First public version.

NgsAdmix: Difference between revisions