what is strobemer? see https://github.com/ksahlin/strobemers

1. this project implement an c++ version of the strobemer s(n,k,w_min,w_max) with small difference (the window will not shrink at the end of sequence).

Features:

• any n>1 is supported.
• randstrobes is supported.
• minstrobes is supported.
• hybirdstrobes is supported.

2. there small benchmark were attached. (Ignore them if you don't care)

The implementation of strobemer

How to use the c++ implemention of strobemer ? See toy_example

• first copy strobemer.h and strobemer.cpp into your project.

• then write your code like :

#include "strobemer.h"

int main() {
     char seq[101]="ATGGGCAGAGTTTGACGTAGTCAATGCTTATGAACGAACGCTCCAATATGAATCAGCTCGTGATTTTTGCTGTAAAAATCGTAGCATACTGTTTGATAAA";
     //strobemer::init(3,13,13,20,strobemer_type::minstrobe); // n=3,k=13,w_min=13,w_max=20,type=minstrobe
     //strobemer::init(3,13,13,21,strobemer_type::hybridstrobe); // n=3,k=13,w_min=13,w_max=20,type=hybridstrobe
     strobemer::init(3,13,13,20,strobemer_type::randstrobe);  // n=3,k=13,w_min=13,w_max=20,type=randstrobe
     int number = 100-strobemer::strobmer_span()+1;
     strobemer * buff = new strobemer[number];
     strobemer::chop_strobemer(seq,100,buff);
     for(int i = 0 ; i< number ; i++ ){
         if(buff[i].valid)
             std::cout<<buff[i].to_string()<<'\n'; // or do whatever you want ...
     }
     delete [] buff;
     return 0;
}

then compile and link it like:

g++  -std=c++11 -c strobemer.cpp -o strobemer.o
g++  -std=c++11 example.cpp strobemer.o -o example

There benchmarks

the test date

• random sequence with length=100,000 nt.
• mutation rates are 0.01, 0.05 and 0.1.
• random mutaion.
• equal probability of sub/ins/del.
• each test run 100 times.

benchmark_SIM-R-match-only

• compare the match number of all kmer/strobemer.
• compare k30, randstrobe(2,15,50) and minstrobe(2,15,50).

Results of the "average match of all kmers/strobemers (%) " for different error rates and different methods:

	0.01	0.05	0.1
Kmer(30)	74.6	22.5	4.7
minstrobe(2,15,15,50)	70.1	17.5	3.2
randstrobe(2,15,15,50)	70.1	17.5	3.2

benchmark_SIM-R-snp30

• compare the match number of all kmer/strobemer snp markers.
• assign one snp for each 1000bp ( total 99 snps for 100,000bp because I drop the snp at the sequence end ).
• compare k30, randstrobe(2,15,50) and minstrobe(2,15,50).
• for k30, one snp generate 30 kmer markers.
• for strobemer(2,15,50), each snp generate 65 strobemer markers.

Results of the "average match number of all snp markers" for different error rates and different methods:

	0.01	0.05	0.1
Kmer(30)	2210	665	143
minstrobe(2,15,15,50)	4511	1100	208
randstrobe(2,15,15,50)	4502	1134	210

Results of the "#detected snp " for different error rates and different methods:

	0.01	0.05	0.1
Kmer(30)	94.5	55.5	18
minstrobe(2,15,15,50)	98	68.5	26.3
randstrobe(2,15,15,50)	99	92.3	52.3

benchmark_SIM-R-snp20

• compare the match number of all kmer/strobemer snp markers.
• assign one snp for each 1000bp ( total 99 snps for 100,000bp because I drop the snp at the sequence end ).
• compare k20, k40, randstrobe(2,10,30) and minstrobe(2,10,30).
• for K20, one snp generate 20 kmer markers.
• for K40, one snp generate 40 kmer markers.
• for strobemer(2,10,30), each snp generate 40 strobemer markers.

Results of the "average match number of all snp markers" for different error rates and different methods:

	0.01	0.05	0.1
Kmer(20)	1631.5	748.93	260.72
Kmer(40)	2682.38	540.22	62.83
minstrobe(2,10,10,30)	3171.38	1296.55	411.71
randstrobe(2,10,10,30)	3164.28	1292.72	417.33

Results of the "#detected snp " for different error rates and different methods:

	0.01	0.05	0.1
Kmer(20)	96.79	71.05	37.79
Kmer(40)	92.81	39.98	8.01
minstrobe(2,10,10,30)	98.54	85.05	51.82
randstrobe(2,10,10,30)	98.99	95.24	73.39

Enjoy ~~