Lower percentage of properly paired sequence

[sunnyvu]

After alignment using Tophat, I use Samtools to check the flag statistics, I realized that the percentage of properly paired sequence is very lower, around 0.01%. Is this normal?

Thanks in advance.

[Auction]

I met such problem before. Then I found the average mate distance (based on mapping to genome using BWA) is significanntly bigger that the number I sent to "-r" parameter in Tophat (based on people who constructed the library). Then I increase the "-r" paramter and get more properly paired reads. You can also try it.

After alignment using Tophat, I use Samtools to check the flag statistics, I realized that the percentage of properly paired sequence is very lower, around 0.01%. Is this normal?

Thanks in advance.

[sunnyvu]

Auction, I followed your suggestion to increase the parameter. I already increased r to 1000, I got more but just about 0.03%. How about your experiment?

I had a question too. Is it reasonable to increase r to 1000 since the expected fragment is only around 200bp.

Thanks.

[Auction]

sunnyvu, my idea is to adjust the "-r" according to the real insert size of your dataset. Therefore you may need decrease in your dataset depends on the situation. I think you can use first use BWA to align your reads to the genome because BWA will automatically estimate the insert size. And you could also estimate it from result SAM file.

Auction, I followed your suggestion to increase the parameter. I already increased r to 1000, I got more but just about 0.03%. How about your experiment?

I had a question too. Is it reasonable to increase r to 1000 since the expected fragment is only around 200bp.

Thanks.