Bridge Amplification error rate and duplicates in NGS

Question

[KirGen]

Hello!

I have 3 questions,can you help me please to understand?

1. Why almost only the pcr step in the library prep is cited as source of errors while bridge amplification not? I have found only one paper that talks about the bridge pcr errors..why? I think that they are both pcr steps so they should have the same error rates,it is correct or not?
2. Why duplicates are creates only during pcr step in library prep and not durind brigde pcr? I haven't found nothing in literature about bridge pcr as source of duplicates..but why? They are both pcr steps so they should have the same probability to create duplicates..Is it false?
3. If I remove pcr duplicate, why I don't loose the massive paralles sequencing power of ngs? I know that the benefit of ngs that allow to reach high specificity in sequencing is to sequence a lot of time the same fragment..but if I remove duplicates from the analysis i don't loose this benefit?

Thank you in advance for your help!

Answer 1

[finswimmer]

Hello,

I think all your question can be solve if you recall what these 3 thing are:

1. A single molecule derived after library preparation that will bind on the flow cell
2. A cluster on the flow cell
3. A read sequence

A single molecule from the library preparation will bind on the flow cell. It gets amplified by bridge amplification and will form a cluster, where all molecule are duplicates and located next to each other. These cluster will give the sequence machine a signal about it's sequence during the sequence process. You will receive exactly one sequence per cluster (in case of paired end read two) which is called a read.

Saying this, it means when you create duplicates during library preparation, these duplicates will form independent cluster that result in multiple read sequences from the same origin. If one of these duplicated molecule have in pcr error, you will see this probably in the read sequence. In contrast if you have pcr errors within a cluster, you have thousands of amplified molecule within the cluster that doesn't. So you will have a background noise of these errors, but usually this doesn't matter.

[KirGen]

Thank you very much for your answer! If I understand well, removing duplicates I don't lose the "power" of NGS because high depth is given by sequencing more and more time the same positions that come from different fragments with different sizes and not from duplicates. But removing duplicates I don't lose coverage? Another question: if I have hybridization probes that are for example 120mer and the gDNA is randomly shared in fragments of 350-400 bp, could I have probes that recognize 100 bp of one fragment and 20 bp in another one or 60bp in one fragment and 60 in another one because the hybridization site has been divided in two fragments? Thank you in advance!!