The project that I am doing now requires me to filter of hg19 sequences from my data set. I have trimmed my read lengths to 50 bp with Qphred = 20. Next, I will be using bowtie2 to align the sequences. I would like to know how can I know if the default -X value in bowtie2 is suitable for the analysis of my data.
If the default -X value is not suitable, can you suggest how can I change the default parameter?
I am new to bioinformatics. Thank you very much for your help.
depends on your data. Is it paired-end?. If it is then ask your lab what was the insert size that they selected for library preparation. The whole point of the -I and -X parameters is to narrow the search for the reads in pair. If the insert was 500, and the length of reads in pair is 100, than after mapping first reads of the pair bowtie will be optimized to look for the mapping of the other read from the pair close to the other end of the insert. It probably won't make that large of the difference if you miss it a bit. Note that I say optimize. By default bowtie 2 is going to look within the whole range of 0-500 bp insert for the mapping of the other mate of the pair. If your insert size was much larger than that 500 bp, then you have a problem and probably a lot of singleton mappings.
If you don't know the insert size of your library at all you can estimate it. Be it either bowtie or bwa mapping, but the resulting BAM file you pass to the PicardTools and use the insert size metrics to estimate that value. Then I guess taking lower and upper confidence intervals should do ok for your -I -X paramters to fine tune the mapping with bowtie2
Also with paired-end you might want to consider collapsing the overlapoing reads with something like adapterremoval. The reads that are overlapping each other will tend to underestimate the insert size of your library
depends on your data. Is it paired-end?. If it is then ask your lab what was the insert size that they selected for library preparation. The whole point of the -I and -X parameters is to narrow the search for the reads in pair. If the insert was 500, and the length of reads in pair is 100, than after mapping first reads of the pair bowtie will be optimized to look for the mapping of the other read from the pair close to the other end of the insert. It probably won't make that large of the difference if you miss it a bit. Note that I say optimize. By default bowtie 2 is going to look within the whole range of 0-500 bp insert for the mapping of the other mate of the pair. If your insert size was much larger than that 500 bp, then you have a problem and probably a lot of singleton mappings.
If you don't know the insert size of your library at all you can estimate it. Be it either bowtie or bwa mapping, but the resulting BAM file you pass to the PicardTools and use the insert size metrics to estimate that value. Then I guess taking lower and upper confidence intervals should do ok for your -I -X paramters to fine tune the mapping with bowtie2
Also with paired-end you might want to consider collapsing the overlapoing reads with something like adapterremoval. The reads that are overlapping each other will tend to underestimate the insert size of your library
Hi Ireneusz,
Thank you very much for your commend. I have read a more about Bowtie because of your suggestion. I understand it more now.
Best regards,
Grace
I see modern day tools do not consider insert size while alignment and quantification of the paired reads. Is not to worth considering?
- Badges
- Science method
- Similar topics
- Biological Science
- Genetics
- Genetic Analysis
- genetic techniques
- Sequencing