3' bias in Illumina sequencing

[subuhikhan]

Hello,

I am working on a plant virus RNA sequencing data (Illumina). I have trimmed my reads from 101bp to around 90bp to improve the quality of each read with the library. I created assembly of my sequence libraries against the reference Viral genome sequence (10kb in size). The mappability of reads is showing bias towards 3'end of the viral genome i.e. more reads are binding towards the 3' end? What can be the possible reason for that?? And how can I reduce the bias at the 3' end?

Thank you
Subuhi

[rskr]

Hello,

I am working on a plant virus RNA sequencing data (Illumina). I have trimmed my reads from 101bp to around 90bp to improve the quality of each read with the library. I created assembly of my sequence libraries against the reference Viral genome sequence (10kb in size). The mappability of reads is showing bias towards 3'end of the viral genome i.e. more reads are binding towards the 3' end? What can be the possible reason for that?? And how can I reduce the bias at the 3' end?

Thank you
Subuhi

In various transcriptome experiments we *may* have observed some strand bias. I don't think it has much to do with the sequencing. If it is true that the bias is there, I think it might be do to the PCR reactions for RNA being incomplete in the first cycle, and this again will depend on the method of amplification for the primers. But if there were a bunch of strand specific template, and a certain fraction were only amplified partially in the first cycle then you would see a bias, since at every iteration of the reaction the sequences wouldn't get longer.

I don't know if this helps, its a hypothesis, but I am interested to see if any one else has similar suspicions.

[subuhikhan]

Hello,

But if we are first fragmenting the RNA and then making its cDNA even then we will see this bias (i.e. the bias caused by RNA primers used for sequencing)?

Thank u
Subuhi

[rskr]

Hello,

But if we are first fragmenting the RNA and then making its cDNA even then we will see this bias (i.e. the bias caused by RNA primers used for sequencing)?

Thank u
Subuhi

Hmm... first fragmenting RNA, then making cDNA. Correct me if I am wrong, but cDNA molecules will have three prime ends on either end of the double stranded DNA. Doesn't sound like a sequencing bias. Your RNA may be enzymatically active and or forming structures, which could potentially behave as a bias perhaps.

[HESmith]

Hmm... first fragmenting RNA, then making cDNA. Correct me if I am wrong, but cDNA molecules will have three prime ends on either end of the double stranded DNA. Doesn't sound like a sequencing bias. Your RNA may be enzymatically active and or forming structures, which could potentially behave as a bias perhaps.

Not wrong, but irrelevant to the poster's question. It's a 10 kbp viral RNA genome, and an excess of reads maps to the 3' end of that genome.

A few possibilities come to mind: 1) the sample is degraded from the 5' end; 2) assuming a polyadenylated plus-strand genome, first strand synthesis was primed with oligo(dT) instead of the more typical oligo(dT) plus random hexamer mix; 3) the sequenced genome is flanked by LTRs, but the reference used for alignment contains only one copy at the 3' end.