Bowtie2 with multiple hits

Conditions

• On the progress of implementing bowtie2, it might question whether there are several hits to the same read ? But, bowtie2 seemed that it only output a single result by default.

• After executing alignments on NGS data with bowtie2 and statistical analyses on alignment results, it would find statistical results seem not coordinate with well-known (or real) findings, especially between computing ones and biological ones.

• The real condition is that one read would map to several hits in high scores between progress of implementing bowtie2 , but bowtie2 would only output one hit, the best one, by default. (more details could be refer to official document [1])

Solutions

• There are two methods (or parameters), -k and -a modes , for executing bowtie2 with results of multiple hits to a read.

  1. -k Number (Number means how many maps should be extracted): For example, "-k 3" means the results would show at most three hits from the database in one read.

  2. -a (All) :

     • Similar with -k mode, -a mode makes bowtie2 not to set limitation of counting for a single read .
     • The mapping score for this hit equals to the sum of each pair mapped with one .
     • The representing is the sorted result by the rule of descending on scores .
     • Only the first hit (the best hit) would not be marked at 9th bit (means 2^8 = 256) in the FLAG field and the other would be set for a reason of "not primary hit".
     • It might cost lots of time .

Commands

• Following is the simple example, paired-end data, to execute the above two parameters, -k n and -a, in order to extract multiple hits to a single read.

  1. -k n mode:

  [ ~ ] $ ./bowtie2-align -x example\index\lambda_virus -1 example\reads\reads_1.fq -2 example\reads\reads_2.fq -k 10 -S lambda.sam
  

  2. -a mode:

  [ ~ ] $ ./bowtie2-align -x example\index\lambda_virus -1 example\reads\reads_1.fq -2 example\reads\reads_2.fq -a -S lambda.sam
  

Results

• Simple results after implementing -a or -k mode: It could easily understand one read might be mapped to several hits and the flag would reveal the information that most of hits are not the primary one. The following image could be separated into two parts, the lower is the primary hit to a read and the upper is the other.

References

1. Official document from developers University of John Hopkins [ Link ]