func calculateD(sample []fwd1.Sequence, length int, ch chan dResult) {
dmatrix := fwd1.GenerateDistanceMatrix(sample)
cmatrix := covs.NewCMatrix(sampleSize, length, dmatrix)
ks, vard := cmatrix.D()
dr := dResult{ks: ks, vard: vard}
ch <- dr
}
func main() {
// create file storing ks and vard
f, err := os.Create(fname + ".csv")
if err != nil {
panic(err)
}
defer f.Close()
ksarray := []float64{} // store ks
vdarray := []float64{} // store VarD
ngarray := []float64{} // store generation number
// do evolution
for i := 0; i < ngen; i++ {
pop.Evolve()
// we make 10 samples and average
ksmean := desc.NewMean()
vdmean := desc.NewMean()
for j := 0; j < sampleTime; j++ {
sample := fwd.Sample(pop.Genomes, sampleSize)
dmatrix := fwd.GenerateDistanceMatrix(sample)
cmatrix := covs.NewCMatrix(sampleSize, pop.Length, dmatrix)
ks, vard := cmatrix.D()
ksmean.Increment(ks)
vdmean.Increment(vard)
}
// write to csv
f.WriteString(fmt.Sprintf("%d,%g,%g\n", pop.NumOfGen, ksmean.GetResult(), vdmean.GetResult()))
if (i+1)%1000 == 0 {
fmt.Println("Generation: ", pop.NumOfGen, ksmean.GetResult(), vdmean.GetResult())
}
// store array for draw
ksarray = append(ksarray, ksmean.GetResult())
vdarray = append(vdarray, vdmean.GetResult())
ngarray = append(ngarray, float64(pop.NumOfGen))
}
// draw
svger := render.NewSVG(fname, 1, 2, 800, 200)
pl := chart.ScatterChart{Title: "KS"}
pl.AddDataPair("KS", ngarray, ksarray, chart.PlotStyleLines, chart.Style{Symbol: '+', SymbolColor: "#0000ff", LineStyle: chart.SolidLine})
svger.Plot(&pl)
pl = chart.ScatterChart{Title: "VarD"}
pl.AddDataPair("VarD", ngarray, vdarray, chart.PlotStyleLines, chart.Style{Symbol: '+', SymbolColor: "#0000ff", LineStyle: chart.SolidLine})
svger.Plot(&pl)
svger.Close()
// save population
jf, err := os.Create(fname + ".json")
if err != nil {
panic(err)
}
defer jf.Close()
b, err := json.Marshal(pop)
if err != nil {
panic(err)
}
jf.Write(b)
}
func main() {
// use all cpus
runtime.GOMAXPROCS(runtime.NumCPU())
// population paramters
length := 1000 // genome length = 1000
mutation := 0.00001 // mutation rate = 1e-5
transfer := 0.0 // transfer rate = 0
fragment := 0 // transferred length = 0
// simulation parameters
samplesize := 100
numofgen := 100000
// population size array
sizearray := []int{100, 1000, 10000}
for _, size := range sizearray {
// population
pop := fwd.NewSeqPop(size, length, mutation, transfer, fragment)
// result files
ksname := fmt.Sprintf("ks_size_%d.csv", size)
ksfile, err := os.Create(ksname)
if err != nil {
panic(err)
}
vdname := fmt.Sprintf("vd_size_%d.csv", size)
vdfile, err := os.Create(vdname)
if err != nil {
panic(err)
}
// info file
infoname := fmt.Sprintf("ks_size_%d.txt", size)
infofile, err := os.Create(infoname)
if err != nil {
panic(err)
}
infofile.WriteString(fmt.Sprintf("Size = %d\n", pop.Size))
infofile.WriteString(fmt.Sprintf("Length = %d\n", pop.Length))
infofile.WriteString(fmt.Sprintf("Mutation = %g\n", pop.Mutation))
// population evolve
for i := 0; i < numofgen; i++ {
pop.Evolve()
sample := fwd.Sample(pop.Genomes, samplesize)
dmatrix := fwd.GenerateDistanceMatrix(sample)
cmatrix := covs.NewCMatrix(samplesize, length, dmatrix)
ks, vd := cmatrix.D()
ksfile.WriteString(fmt.Sprintf("%g\n", ks))
vdfile.WriteString(fmt.Sprintf("%g\n", vd))
}
// close files
ksfile.Close()
vdfile.Close()
infofile.Close()
}
}
