########################################################################
###Read fastq file and barcode file
all_seqs<-read.fastq('m:/pc/Desktop/Mikkel/pacBio/Mikkel_PCR_pool/Mikkel_PCR_pool/MikkelPool_reads_of_insert_min3passes.fastq')
length(all_seqs)
hist(unlist(lapply(all_seqs,length)),breaks=100)
barcodes<-read.table('m:/pc/Desktop/Mikkel/pacBio/PCR_control_study/BarcodesC2.txt',header=F)
#filter on length
all_seqs<-all_seqs[-which(unlist(lapply(all_seqs,length))<300)]
all_seqs<-all_seqs[-which(unlist(lapply(all_seqs,length))>400)]
length(all_seqs)

###demultiplex reads
gogo<-demultiplex(barcodes,all_seqs,mm=0)
length(gogo)
names(gogo)
par(las=2)
barplot(unlist(lapply(gogo,length)))
mean(unlist(lapply(gogo,length)))
sd(unlist(lapply(gogo,length)))

#nr. unassigned sequences
length(all_seqs)-sum(unlist(lapply(gogo,length)))

###Extract microsat sequences
#pstI digest control.
#x20 tetra
pstI_4x20<-extr.mi.sat('tccaagcgagag','tcatgtgataaa',gogo[[1]])
length(pstI_4x20)
pstI_4x20_l<-unlist(lapply(pstI_4x20,length))


#x10 tetra
pstI_4x10<-extr.mi.sat('aactaaagtaca','gggttgaccttc',gogo[[1]])
length(pstI_4x10)
summary(unlist(lapply(pstI_4x10,length)))
barplot(table(unlist(lapply(pstI_4x10,length))))
pstI_4x10_l<-unlist(lapply(pstI_4x10,length))

#ecoRI digest control.
#x30 dinuc
#10cyc241
ecoRI_2x30<-extr.mi.sat('taaagttctatc','acccttaatgtt',gogo[[2]])
length(ecoRI_2x30)
summary(unlist(lapply(ecoRI_2x30,length)))
barplot(table(unlist(lapply(ecoRI_2x30,length))))
ecoRI_2x30_l<-unlist(lapply(ecoRI_2x30,length))

#x10 tetra
ecoRI_4x10<-extr.mi.sat('atggactctaca','aagctataaaga',gogo[[2]])
length(ecoRI_4x10)
summary(unlist(lapply(ecoRI_4x10,length)))
barplot(table(unlist(lapply(ecoRI_4x10,length))))
ecoRI_4x10_l<-unlist(lapply(ecoRI_4x10,length))

#10cyc131
#x20 tetra
c10_131_4x20<-extr.mi.sat('tccaagcgagag','tcatgtgataaa',gogo[[3]])
length(c10_131_4x20)
summary(unlist(lapply(c10_131_4x20,length)))
barplot(table(unlist(lapply(c10_131_4x20,length))))
c10_131_4x20_l<-unlist(lapply(c10_131_4x20,length))

#x10 tetra
c10_131_4x10<-extr.mi.sat('aactaaagtaca','gggttgaccttc',gogo[[3]])
length(c10_131_4x10)
summary(unlist(lapply(c10_131_4x10,length)))
barplot(table(unlist(lapply(c10_131_4x10,length))))
c10_131_4x10_l<-unlist(lapply(c10_131_4x10,length))

#10cyc132
#x20 tetra
c10_132_4x20<-extr.mi.sat('tccaagcgagag','tcatgtgataaa',gogo[[5]])
length(c10_132_4x20)
summary(unlist(lapply(c10_132_4x20,length)))
barplot(table(unlist(lapply(c10_132_4x20,length))))
c10_132_4x20_l<-unlist(lapply(c10_132_4x20,length))

#x10 tetra
c10_132_4x10<-extr.mi.sat('aactaaagtaca','gggttgaccttc',gogo[[5]])
length(c10_132_4x10)
summary(unlist(lapply(c10_132_4x10,length)))
barplot(table(unlist(lapply(c10_132_4x10,length))))
c10_132_4x10_l<-unlist(lapply(c10_132_4x10,length))

###################
#10cyc241
c10_241_2x30<-extr.mi.sat('taaagttctatc','acccttaatgtt',gogo[[4]])
length(c10_241_2x30)
summary(unlist(lapply(c10_241_2x30,length)))
barplot(table(unlist(lapply(c10_241_2x30,length))))
c10_241_2x30_l<-unlist(lapply(c10_241_2x30,length))

c10_241_4x10<-extr.mi.sat('atggactctaca','aagctataaaga',gogo[[4]])
length(c10_241_4x10)
summary(unlist(lapply(c10_241_4x10,length)))
barplot(table(unlist(lapply(c10_241_4x10,length))))
c10_241_4x10_l<-unlist(lapply(c10_241_4x10,length))

#10cyc242
c10_242_2x30<-extr.mi.sat('taaagttctatc','acccttaatgtt',gogo[[6]])
length(c10_242_2x30)
summary(unlist(lapply(c10_242_2x30,length)))
barplot(table(unlist(lapply(c10_242_2x30,length))))
c10_242_2x30_l<-unlist(lapply(c10_242_2x30,length))

c10_242_4x10<-extr.mi.sat('atggactctaca','aagctataaaga',gogo[[6]])
length(c10_242_4x10)
summary(unlist(lapply(c10_242_4x10,length)))
barplot(table(unlist(lapply(c10_242_4x10,length))))
c10_242_4x10_l<-unlist(lapply(c10_242_4x10,length))

##########################
#20cyc131
#x20 tetra
c20_131_4x20<-extr.mi.sat('tccaagcgagag','tcatgtgataaa',gogo[[7]])
length(c20_131_4x20)
summary(unlist(lapply(c20_131_4x20,length)))
barplot(table(unlist(lapply(c20_131_4x20,length))))
c20_131_4x20_l<-unlist(lapply(c20_131_4x20,length))

#x10 tetra
c20_131_4x10<-extr.mi.sat('aactaaagtaca','gggttgaccttc',gogo[[7]])
length(c20_131_4x10)
summary(unlist(lapply(c20_131_4x10,length)))
barplot(table(unlist(lapply(c20_131_4x10,length))))
c20_131_4x10_l<-unlist(lapply(c20_131_4x10,length))

#20cyc132
#x20 tetra
c20_132_4x20<-extr.mi.sat('tccaagcgagag','tcatgtgataaa',gogo[[9]])
length(c20_132_4x20)
summary(unlist(lapply(c20_132_4x20,length)))
barplot(table(unlist(lapply(c20_132_4x20,length))))
c20_132_4x20_l<-unlist(lapply(c20_132_4x20,length))

#x10 tetra
c20_132_4x10<-extr.mi.sat('aactaaagtaca','gggttgaccttc',gogo[[9]])
length(c20_132_4x10)
summary(unlist(lapply(c20_132_4x10,length)))
barplot(table(unlist(lapply(c20_132_4x10,length))))
c20_132_4x10_l<-unlist(lapply(c20_132_4x10,length))

############################
#20cyc241
#2x30
c20_241_2x30<-extr.mi.sat('taaagttctatc','acccttaatgtt',gogo[[8]])
length(c20_241_2x30)
summary(unlist(lapply(c20_241_2x30,length)))
barplot(table(unlist(lapply(c20_241_2x30,length))))
c20_241_2x30_l<-unlist(lapply(c20_241_2x30,length))

#4x10
c20_241_4x10<-extr.mi.sat('atggactctaca','aagctataaaga',gogo[[8]])
length(c20_241_4x10)
summary(unlist(lapply(c20_241_4x10,length)))
barplot(table(unlist(lapply(c20_241_4x10,length))))
c20_241_4x10_l<-unlist(lapply(c20_241_4x10,length))

#20cyc242
#2x30
c20_242_2x30<-extr.mi.sat('taaagttctatc','acccttaatgtt',gogo[[10]])
length(c20_242_2x30)
summary(unlist(lapply(c20_242_2x30,length)))
barplot(table(unlist(lapply(c20_242_2x30,length))))
c20_242_2x30_l<-unlist(lapply(c20_242_2x30,length))

c20_242_4x10<-extr.mi.sat('atggactctaca','aagctataaaga',gogo[[10]])
length(c20_242_4x10)
summary(unlist(lapply(c20_242_4x10,length)))
barplot(table(unlist(lapply(c20_242_4x10,length))))
c20_242_4x10_l<-unlist(lapply(c20_242_4x10,length))

####################################
#30cyc131
#x20 tetra
c30_131_4x20<-extr.mi.sat('tccaagcgagag','tcatgtgataaa',gogo[[11]])
length(c30_131_4x20)
summary(unlist(lapply(c30_131_4x20,length)))
barplot(table(unlist(lapply(c30_131_4x20,length))))
c30_131_4x20_l<-unlist(lapply(c30_131_4x20,length))

#x10 tetra
c30_131_4x10<-extr.mi.sat('aactaaagtaca','gggttgaccttc',gogo[[11]])
length(c30_131_4x10)
summary(unlist(lapply(c30_131_4x10,length)))
barplot(table(unlist(lapply(c30_131_4x10,length))))
c30_131_4x10_l<-unlist(lapply(c30_131_4x10,length))

#30cyc132
#x20 tetra
c30_132_4x20<-extr.mi.sat('tccaagcgagag','tcatgtgataaa',gogo[[13]])
length(c30_132_4x20)
summary(unlist(lapply(c30_132_4x20,length)))
barplot(table(unlist(lapply(c30_132_4x20,length))))
c30_132_4x20_l<-unlist(lapply(c30_132_4x20,length))

#x10 tetra
c30_132_4x10<-extr.mi.sat('aactaaagtaca','gggttgaccttc',gogo[[13]])
length(c30_132_4x10)
summary(unlist(lapply(c30_132_4x10,length)))
barplot(table(unlist(lapply(c30_132_4x10,length))))
c30_132_4x10_l<-unlist(lapply(c30_132_4x10,length))

###############################
#30cyc241
c30_241_2x30<-extr.mi.sat('taaagttctatc','acccttaatgtt',gogo[[12]])
length(c30_241_2x30)
summary(unlist(lapply(c30_241_2x30,length)))
barplot(table(unlist(lapply(c30_241_2x30,length))))
c30_241_2x30_l<-unlist(lapply(c30_241_2x30,length))

c30_241_4x10<-extr.mi.sat('atggactctaca','aagctataaaga',gogo[[12]])
length(c30_241_4x10)
summary(unlist(lapply(c30_241_4x10,length)))
barplot(table(unlist(lapply(c30_241_4x10,length))))
c30_241_4x10_l<-unlist(lapply(c30_241_4x10,length))

#30cyc242
c30_242_2x30<-extr.mi.sat('taaagttctatc','acccttaatgtt',gogo[[14]])
length(c30_242_2x30)
summary(unlist(lapply(c30_242_2x30,length)))
barplot(table(unlist(lapply(c30_242_2x30,length))))
c30_242_2x30_l<-unlist(lapply(c30_242_2x30,length))

c30_242_4x10<-extr.mi.sat('atggactctaca','aagctataaaga',gogo[[14]])
length(c30_242_4x10)
summary(unlist(lapply(c30_242_4x10,length)))
barplot(table(unlist(lapply(c30_242_4x10,length))))
c30_242_4x10_l<-unlist(lapply(c30_242_4x10,length))

#####################################
#40cyc131
#x20 tetra
c40_131_4x20<-extr.mi.sat('tccaagcgagag','tcatgtgataaa',gogo[[15]])
length(c40_131_4x20)
summary(unlist(lapply(c40_131_4x20,length)))
barplot(table(unlist(lapply(c40_131_4x20,length))))
c40_131_4x20_l<-unlist(lapply(c40_131_4x20,length))

#x10 tetra
c40_131_4x10<-extr.mi.sat('aactaaagtaca','gggttgaccttc',gogo[[15]])
length(c40_131_4x10)
summary(unlist(lapply(c40_131_4x10,length)))
barplot(table(unlist(lapply(c40_131_4x10,length))))
c40_131_4x10_l<-unlist(lapply(c40_131_4x10,length))

#40cyc132
#x20 tetra

c40_132_4x20<-extr.mi.sat('tccaagcgagag','tcatgtgataaa',gogo[[17]])
length(c40_132_4x20)
summary(unlist(lapply(c40_132_4x20,length)))
barplot(table(unlist(lapply(c40_132_4x20,length))))
c40_132_4x20_l<-unlist(lapply(c40_132_4x20,length))

#x10 tetra
c40_132_4x10<-extr.mi.sat('aactaaagtaca','gggttgaccttc',gogo[[17]])
length(c40_132_4x10)
summary(unlist(lapply(c40_132_4x10,length)))
barplot(table(unlist(lapply(c40_132_4x10,length))))
c40_132_4x10_l<-unlist(lapply(c40_132_4x10,length))

######################################
#40cyc241
#x30 dinuc
c40_241_2x30<-extr.mi.sat('taaagttctatc','acccttaatgtt',gogo[[16]])
length(c40_241_2x30)
summary(unlist(lapply(c40_241_2x30,length)))
barplot(table(unlist(lapply(c40_241_2x30,length))))
c40_241_2x30_l<-unlist(lapply(c40_241_2x30,length))

#x10 tetra
c40_241_4x10<-extr.mi.sat('atggactctaca','aagctataaaga',gogo[[16]])
length(c40_241_4x10)
summary(unlist(lapply(c40_241_4x10,length)))
barplot(table(unlist(lapply(c40_241_4x10,length))))
c40_241_4x10_l<-unlist(lapply(c40_241_4x10,length))

#40cyc242
#x30 dinuc
c40_242_2x30<-extr.mi.sat('taaagttctatc','acccttaatgtt',gogo[[18]])
length(c40_242_2x30)
summary(unlist(lapply(c40_242_2x30,length)))
barplot(table(unlist(lapply(c40_242_2x30,length))))
c40_242_2x30_l<-unlist(lapply(c40_242_2x30,length))

#x10 tetra
c40_242_4x10<-extr.mi.sat('atggactctaca','aagctataaaga',gogo[[18]])
length(c40_242_4x10)
summary(unlist(lapply(c40_242_4x10,length)))
barplot(table(unlist(lapply(c40_242_4x10,length))))
c40_242_4x10_l<-unlist(lapply(c40_242_4x10,length))

#####################################################################
###extract internal spacer sequences
###No pcr controls
pstI_spacer<-extr.spacer('tcatgtgataaa','aactaaagtaca',gogo[[1]])
length(pstI_spacer)
summary(unlist(lapply(pstI_spacer,length)))
pstI_spacer_l<-unlist(lapply(pstI_spacer,length))
barplot(table(pstI_spacer_l))
table(pstI_spacer_l)/length(pstI_spacer)
pstI_s_id<-indel.rate(pstI_spacer_l,140)

ecoRI_spacer<-extr.spacer('acccttaatgtt','atggactctaca',gogo[[2]])
length(ecoRI_spacer)
summary(unlist(lapply(ecoRI_spacer,length)))
ecoRI_spacer_l<-unlist(lapply(ecoRI_spacer,length))
barplot(table(ecoRI_spacer_l))
ecoRI_s_id<-indel.rate(ecoRI_spacer_l,133)

###############
#10cyc131
c10_131_spacer<-extr.spacer('tcatgtgataaa','aactaaagtaca',gogo[[3]])
length(c10_131_spacer)
summary(unlist(lapply(c10_131_spacer,length)))
c10_131_spacer_l<-unlist(lapply(c10_131_spacer,length))
barplot(table(c10_131_spacer_l))
c10_131_s_id<-indel.rate(c10_131_spacer_l,140)

#10cyc132
c10_132_spacer<-extr.spacer('tcatgtgataaa','aactaaagtaca',gogo[[5]])
length(c10_132_spacer)
summary(unlist(lapply(c10_132_spacer,length)))
c10_132_spacer_l<-unlist(lapply(c10_132_spacer,length))
barplot(table(c10_132_spacer_l))
c10_132_s_id<-indel.rate(c10_132_spacer_l,140)

###################
#10cyc241
c10_241_spacer<-extr.spacer('acccttaatgtt','atggactctaca',gogo[[4]])
length(c10_241_spacer)
summary(unlist(lapply(c10_241_spacer,length)))
c10_241_spacer_l<-unlist(lapply(c10_241_spacer,length))
barplot(table(c10_241_spacer_l))
c10_241_s_id<-indel.rate(c10_241_spacer_l,133)

#10cyc242
c10_242_spacer<-extr.spacer('acccttaatgtt','atggactctaca',gogo[[6]])
length(c10_242_spacer)
summary(unlist(lapply(c10_242_spacer,length)))
c10_242_spacer_l<-unlist(lapply(c10_242_spacer,length))
barplot(table(c10_242_spacer_l))
c10_242_s_id<-indel.rate(c10_242_spacer_l,133)

##########################
#20cyc131
c20_131_spacer<-extr.spacer('tcatgtgataaa','aactaaagtaca',gogo[[7]])
length(c20_131_spacer)
summary(unlist(lapply(c20_131_spacer,length)))
c20_131_spacer_l<-unlist(lapply(c20_131_spacer,length))
barplot(table(c20_131_spacer_l))
c20_131_s_id<-indel.rate(c20_131_spacer_l,140)

#20cyc132
c20_132_spacer<-extr.spacer('tcatgtgataaa','aactaaagtaca',gogo[[9]])
length(c20_132_spacer)
summary(unlist(lapply(c20_132_spacer,length)))
c20_132_spacer_l<-unlist(lapply(c20_132_spacer,length))
barplot(table(c20_132_spacer_l))
c20_132_s_id<-indel.rate(c20_132_spacer_l,140)

############################
#20cyc241
c20_241_spacer<-extr.spacer('acccttaatgtt','atggactctaca',gogo[[8]])
length(c20_241_spacer)
summary(unlist(lapply(c20_241_spacer,length)))
c20_241_spacer_l<-unlist(lapply(c20_241_spacer,length))
barplot(table(c20_241_spacer_l))
c20_241_s_id<-indel.rate(c20_241_spacer_l,133)

#20cyc242
c20_242_spacer<-extr.spacer('acccttaatgtt','atggactctaca',gogo[[10]])
length(c20_242_spacer)
summary(unlist(lapply(c20_242_spacer,length)))
c20_242_spacer_l<-unlist(lapply(c20_242_spacer,length))
barplot(table(c20_242_spacer_l))
c20_242_s_id<-indel.rate(c20_242_spacer_l,133)

####################################
#30cyc131
c30_131_spacer<-extr.spacer('tcatgtgataaa','aactaaagtaca',gogo[[11]])
length(c30_131_spacer)
summary(unlist(lapply(c30_131_spacer,length)))
c30_131_spacer_l<-unlist(lapply(c30_131_spacer,length))
barplot(table(c30_131_spacer_l))
c30_131_s_id<-indel.rate(c30_131_spacer_l,140)

#30cyc132
c30_132_spacer<-extr.spacer('tcatgtgataaa','aactaaagtaca',gogo[[13]])
length(c30_132_spacer)
summary(unlist(lapply(c30_132_spacer,length)))
c30_132_spacer_l<-unlist(lapply(c30_132_spacer,length))
barplot(table(c30_132_spacer_l))
c30_132_s_id<-indel.rate(c30_132_spacer_l,140)

###############################
#30cyc241
c30_241_spacer<-extr.spacer('acccttaatgtt','atggactctaca',gogo[[12]])
length(c30_241_spacer)
summary(unlist(lapply(c30_241_spacer,length)))
c30_241_spacer_l<-unlist(lapply(c30_241_spacer,length))
barplot(table(c30_241_spacer_l))
c30_241_s_id<-indel.rate(c30_241_spacer_l,133)

#30cyc242
c30_242_spacer<-extr.spacer('acccttaatgtt','atggactctaca',gogo[[14]])
length(c30_242_spacer)
summary(unlist(lapply(c30_242_spacer,length)))
c30_242_spacer_l<-unlist(lapply(c30_242_spacer,length))
barplot(table(c30_242_spacer_l))
c30_242_s_id<-indel.rate(c30_242_spacer_l,133)

#####################################
#40cyc131
c40_131_spacer<-extr.spacer('tcatgtgataaa','aactaaagtaca',gogo[[15]])
length(c40_131_spacer)
summary(unlist(lapply(c40_131_spacer,length)))
c40_131_spacer_l<-unlist(lapply(c40_131_spacer,length))
barplot(table(c40_131_spacer_l))
c40_131_s_id<-indel.rate(c40_131_spacer_l,140)

#40cyc132
c40_132_spacer<-extr.spacer('tcatgtgataaa','aactaaagtaca',gogo[[17]])
length(c40_132_spacer)
summary(unlist(lapply(c40_132_spacer,length)))
c40_132_spacer_l<-unlist(lapply(c40_132_spacer,length))
barplot(table(c40_132_spacer_l))
c40_132_s_id<-indel.rate(c40_132_spacer_l,140)

######################################
#40cyc241
c40_241_spacer<-extr.spacer('acccttaatgtt','atggactctaca',gogo[[16]])
length(c40_241_spacer)
summary(unlist(lapply(c40_241_spacer,length)))
c40_241_spacer_l<-unlist(lapply(c40_241_spacer,length))
barplot(table(c40_241_spacer_l))
c40_241_s_id<-indel.rate(c40_241_spacer_l,133)

#40cyc242
c40_242_spacer<-extr.spacer('acccttaatgtt','atggactctaca',gogo[[18]])
length(c40_242_spacer)
summary(unlist(lapply(c40_242_spacer,length)))
c40_242_spacer_l<-unlist(lapply(c40_242_spacer,length))
barplot(table(c40_242_spacer_l))
c40_242_s_id<-indel.rate(c40_242_spacer_l,133)