Hi all,
I have a few questions about next generation sequencing and the generation of the reads. I did not understand some parts of the process..In particular, are there more reads generated for each sequence read or just one ?
And if there are many reads for each sequence, how does the sequencing process determine which read is the most reliable to generate the contigs? Does it perform a consensus between these reads?
How can you be sure that the variations in the sequence you see (in comparison with a known genome) are real mutations or come from an error in the sequencing?
Thank you very much
Silvia, There is not a simple answer to your question, because the answer can depend on how you prepared your samples. However, in a typical sheared library preparation, you have to remember that there are some PCR steps (and ligations) involved in the library preparation before they go on the chip. Thus, point mutations and apparent translocations can occur in individual molecules early in the process and be amplified by the later PCR steps. Thus, there is expected to be some level of PCR duplicates that represent the same original molecule that was amplified. Normally this is a low level and there is software to help eliminate these duplicates, although it does not work perfectly. It is widely accepted that you should not accept a mutation that occurs only in one sequence read. Ideally, you should have any mutation 3 or more times from reads that were clearly not PCR duplicates (this is best judged by reads that do not start and stop at the same place). This is hard to judge if you are carrying out NGS on molecules that start as PCR products. NGS will have 1-2% errors and therefore any single read is unreliable. Mistakes are much more frequent deeper into the reads.
Thank you Prescott for your answer. So, ideally what you could do is a consensus for the different reads you have to see if the mutation is realistic or not.. (?)Sorry, I am trying to get familiar with the procedure, it is new for me.
Thanks
Depending on the DNA source and the methods used, you should have anywhere from a few to as many as hundreds of thousands of reads per sequenced region. If the template was a small plasmid or phage or DNA virus you would expect more reads per region than if the template was a eukaryotic genome, for example. If the template is one cloned bacteria (colony picked from plate) you expect all reads to be identical and the consensus is good because errors can be from PCR or sequencing error. If the template was a diploid eukaryote, real SNPs should be present in roughly 50:50 ratio (except of X and Y unique regions in mammals, and mitochondria genome).
NGS can be done dozens of different ways on any given sample, and samples can be pretty much anything that would contain DNA or RNA. So it is important to know something about the whole process, before deciding how to analyze the data.
- Badges
- Science method
- Similar topics
- Biological Science
- Genetics
- Genetic Analysis
- genetic techniques
- Sequencing