Home

###Descriptions

1. Population model assumptions

   The population in each generation evolves following standard Wright Fisher model without mutation and selection. That is, randomly sample two individuals from the population, and randomly choose one of the chromosomes in each individual, pair them and recombine to form a new chromosome pair for next generation. Repeat this process until sampled N chromosome pair. Here N denotes population size in specific generation.

   Recombination is modeled as a Poisson Process along the chromosome with rate 1.

2. What the admixture model simulator can do.

   Here implemented a very flexible admixture model simulator, in which:

   1). Can take arbitrary number of ancestral populations;

   2). Can take arbitrary wave of population admixture;

   3). Population size can be changed generation by generation;

   4). And admixture proportions can also be changed generation by generation.

###Get started

1. Requirements

   To run the simulator, java 1.6 or upper is required.

2. Obtain Admsimulator

   AdmSimulator can be either downloaded directly from the release from the URL: https://github.com/xyang619/AdmSimulator/releases or cloned from github repository by

     git clone https://github.com/xyang619/AdmSimulator.git
   

3. Get command line help

    java -jar AdmSimulator.jar -h  or java -jar AdmSimulator.jar --help
   

4. Run it with toy data

    java -jar AdmSimulator.jar --gen 10 --nanc 2 --len 1.0 --file toy.par --samp 20 --prefix test --output test
   

###Input and output

1. Input files

   1). model description file

   firstly, in model description file, set up initial number of haplotypes to be sampled for ancestral populations

   secondly, set up the population size and ancestral proportions for each generation in one line

   Note: anything follow "#" was treated as comments

   Here is a complete example:

    #set up number of haplotypes in each ancestral population to be sampled from
    10	 10
    // #indicate start of population size and ancestral proportions for each generation
    100	0.5	0.5 #init population, two ancestral population, each contribute 50%
    100	0	0
    100	0.1	0 #second wave admixture, with 10% gene flow from ancestral population 1
    100	0	0
    100	0	0
    100	0	0.2 #third wave admixture, with 20% gene flow from ancestral population 2
    100	0	0
    150	0	0 #population increase to 150
    100	0	0 #population size decrease to 100
    100	0	0
   

   It's quite simple to implemented HI, GA or CGFR or CGFD models as described in Jin Wenfei et al (2012) AJHG.

   HI model:

    10 10
    //
    100	0.7	0.3 #init population, two ancestral population, contribute 70% and 30%
    100	0	0
    ......
    100	0	0
   

   GA model:

    10 10
    //
    100	0.7	0.3 #init population, two ancestral population, contribute 70% and 30%
    100	0.1	0.1
    ......
    100	0.1	0.1
   

   CGFD model:

    10 10
    //
    100	0.7	0.3 #init population, two ancestral population, contribute 70% and 30%
    100	0.1	0
    ......
    100	0.1	0
   

   CGFR model:

    10 10
    //
    100	0.7	0.3 #init population, two ancestral population, contribute 70% and 30%
    100	0	0.1
    ......
    100	0	0.1
   

   2). Map file

   The genetic positions for each marker are given in Morgan, one line per marker.

   Here is an example:

    0.00097100
    0.00238066
    0.00367538
    ......
   

   3). Haplotype file

   The haplotypes of ancestral populations to be sampled from are combined in one file, one haplotype per line. The first n1 lines correspond to haplotypes for first ancestral population, second n2 lines correspond to haplotypes for second ancestral population and so on. In which the number of ancestral populations and the number of haplotypes for each ancestral population ((n1, n2 and so on) are given in model description file.

   Here is an example:

    1011000000100100001000000010110100010101000011010111000000000010000000011101100000000101000010010010
    0010000010000100001000000010110101010011000001011111001000000010000000011001100000010101000010110010
    0010000010000100001000000010110101010001010001010111000000100011000000011001110010000101000010010010
    0010000010000100001010000010110101010001000001010111000000100010000000011001110010000101000010010010
    ......
    0100000000000110100000010000000100010000001000000001010001000010000101001011000100001100000111000100
   

2. Output file

   1). Admixed haplotypes

   Record the haplotypes from the individuals sampled from the admixed population, format is the same as input haplotype file

   2). Segment file

   In which record the start, end and which ancestral that segment comes from. Each line corresponds to one chromosomal segment.

   Here is an example:

     0.00000000      0.07785695      2
     0.07785695      0.30178126      1
     ......
     0.30178126      0.41594482      2
   

###Complete arguments list

-h/--help	print help message
-g/--gen	generation since admixture
-k/--nanc	number of ancestral populations
-l/--len	length of chromosome to be simulated
-f/--file	model description parameter file
-n/--samp	number of individuals sampled
-p/--prefix	prefix of input file
-o/--output	prefix of output file
-s/--seed	seed of random generator

###Couple with MS It's very easy to couple with a coalescent simulator MS. For example, use MS to simulate two ancestral populations, whose Ne remains constant, i.e. Ne=5000, split 4000 generations ago, command as below:

    ms 200 1 -t 2000 -r 2000 10000000 -I 2 100 100 -ej 0.2 2 1 -p 8 > simAnc.txt

Then just simply convert into the files needed in the admixture model simulator:

    python convert.py simAnc.txt simAnc 200

It will produce the map file and the ancestral haplotype file:

    simAnc.map simAnc.hap

Afterwards, the simulated ancestral haplotypes could be used in the admixture model simulator, for example:

    java -Xmx2g -jar AdmSimulator.jar -g 20 -k 2 -l 2 -f sim1.par -n 100 -p simAnc -o sim1

Notes: the software is free and open source, users are at their own risk and without any guaranteed.