Why do some Ns have higher quality values than other Ns?

[lcollado]

Hello,

I was recently discussing Illumina FASTQ files with someone and the following question came up:
Why do some Ns have higher quality values than other Ns? Shouldn't they all be the same?

As I understand it, an N pops up when the intensity between 2 bases is so close that the base caller decides that it cannot decide which base was sequenced.

As for higher quality Ns, another part of the base calling depends on how isolated the cluster is. I guess that a higher quality N means that the cluster was more isolated (at least non-overlapping with others) than an N with a low quality. If the cluster is more isolated then the intensity readings should be higher, right?

Just as a cultural fact, we would like to know the correct answer ^_^

Greetings,
Leonardo

[Torst]

I was recently discussing Illumina FASTQ files with someone and the following question came up:
Why do some Ns have higher quality values than other Ns? Shouldn't they all be the same?
As I understand it, an N pops up when the intensity between 2 bases is so close that the base caller decides that it cannot decide which base was sequenced.

In fact, I've seen cases of N having higher quality than some non-N bases in the same read! I suspect it has to do with various factors, but here are two extreme examples which could explain it:

1. No signals at all, N gets called with very low quality.

2. Bi-modal signal, two strong peaks (A and T say), but as Illumina won't about the IUPAC code for "A or T" by default, it has to output N, but because the two signals were clear (?) it calculates a higher quality.

[lcollado]

In fact, I've seen cases of N having higher quality than some non-N bases in the same read!

I have seen the same thing!

I hadn't thought of your 1st point and it makes complete sense Just so others understand it, say that at cycle 1 you have a cluster at tile: 1, X: 2, Y: 3. However, maybe at cycle 30 the same cluster (tile: 1, X: 2, Y: 3) did not produce above-background intensity readings. Hence, the base caller decides that it's an N.

As for your second point, I agree that such scenario could happen. However, even if it's bimodal, the difference between the two modes should be somewhat small. Otherwise I guess that the base caller chooses the highest peak.


Torst, It's always a pleasure to read your replies

Greetings,
Leonardo