Relationship between sequencing lane and ngs dataset

Question

I am fairly new to NGS data analysis and I am struggling to understand the exact relationship between a sequencing lane and an NGS dataset. I should add I don't work in the lab, I only do bioinformatics.

I understand the basics of how to make NGS libraries (very briefly: shearing, ligation of adapters, optional PCR, ligation to solid surface, following by sequencing-by-synthesis or other method). What I am struggling with is the visualisation of what happens once you have made your libraries. My questions are:

• If you have used adapters containing barcode sequences (that identify each patient), you would presumably end up with one single tube with your libraries in it right?

• If so, you then "put" the contents of that tube along with DNA polymerase and nucleotides into one single lane of a flow cell? Your provider would then give you raw FASTQ files that have been identified using this barcode? So you would effectively received one (or 2, depending if it's single/paired-end sequencing) FASTQ file per patient right?

• If you do not use adapters with barcodes, does that mean you would have to use a sequencing lane per patient? If so, that would vastly increase the price you pay your provider, would it not?

  Apologies if these are basic questions.

Answer 1

If you have used adapters containing barcode sequences (that identify each patient), you would presumably end up with one single tube with your libraries in it right?

Correct, you have to prep each sample into a library.

If so, you then "put" the contents of that tube along with DNA polymerase and nucleotides into one single lane of a flow cell?

You could, or you could pool it with other samples (assuming they have compatible indices). As @swbarnes2 mentioned, you don't see a lot of samples run on a single lane these days. Sequencing centers like to split things across lanes in case on lane fails for some reason or there are bubbles in the flow cell or other technical issues.

Your provider would then give you raw FASTQ files that have been identified using this barcode?

Typically, yes, and if your sample was run on multiple lanes they'll often be merged for you automatically. This isn't always the case. I've seen a number of people receive samples either still split by lane (annoying, but you can just cat things together) or, much worse, not demultiplexed at all. In the latter case, people often receive 3 (or 4) fastq files per lane, one for each read and another for each index read. It's rare that sequencing facilities do this, but it happens.

If you do not use adapters with barcodes, does that mean you would have to use a sequencing lane per patient? If so, that would vastly increase the price you pay your provider, would it not?

You could sample a sequence per lane. Whether that makes economic sense is up to you to decide.

Answer 2

It is worth noting that the Illumina NovaSeq and NextSeq address all of the lanes of the flow cell from a single loading point (by default). So on those instruments, you end up with all libraries loaded across all lanes. Therefore for each index you will have L001, L002 (+ L003 and L004 if appropriate) FASTQ files (and R1/R2, again if appropriate).

As @Devon mentions, these files can be cat-ted together, but for some processes modelling the lane effect does matter (BQSR, for example).

Answer 3

Indicies matter, lanes do not, unless you have a screw up on the instrument that messes up some lanes but not all. I'm not sure if anyone skips indices just because they put only one sample on a lane nowadays...it's valuable to be able to use the indices to catch sample mix-ups.