Base intensities

[share]

Hi

Do any one have information regarding the following

1. Acceptable intensity rages (Low and High) for all four bases for GAIIx.
2. Number of clusters/mm2 for optimal data output.
3. Whether they(1,2) differ for flow cell versions.

Thanks in advance
Share

[csc]

1. Acceptable intensity ranges will vary a bit from machine to machine and run to run.

2. 700-800K clusters/mm2 generally when using SCS2.8/RTA1.8 (or 2.9/1.9)

3. The above should apply for v4/v5 flow cells in general.

[share]

Hi csc

If it differs run to run then how would one know whether the run is good. when all the systems have same optical then why don't they have the same intensity cutoffs.

[csc]

Hi share,

Good question. The brief answer is: for one, not all GAs have the exact same parts in them (filters, mode scramblers are pieces I know can vary). Second, every GA is tuned/aligned by hand mostly by different people (i.e. x-y stage flatness, alignment of laser footprint). Third, laser performance changes over time. Those are some hardware things. Next, raw intensities are governed by a number of things. Different cluster densities will give different raw intensities. Chemistry is a big difference. Does the polymerase incorporate identically from run to run? No. This will depend on the activity of the polymerase in the sequencing phase (how fresh it is, how was it stored, temp fluctuations, etc.). The template sequence also matters during cluster amplification (i.e. cluster amplification efficiency is affected by base composition). Layer enzymatic amplification efficiency on top of that. Differential cluster generation will lead to different intensities. Basically compound multiple factors together and you get a broad range.

I have seen lanes with raw first cycle intensities of 300 and 1500 give similar quality data (at least in short reads). Illumina has even told me (and I believe them) that raw intensity is not the best metric to determine if things are going well. There are many things to look at. What is the focus quality like? What is the % intensity after cycle 20? What are the phasing/pre-phasing numbers looking like? How many reads are passing filter? How does the PhiX data look (IVC, error rate, etc.)? How does the calibration data look? If you had to choose a single number, then you should probably go with quality score as it is a composite of numerous factors (but you won't readily see it until after cycle 25).

Do you have a specific problem or issue you are trying to troubleshoot? In the end, by performing a lot of runs you get better at knowing what combination of metrics is indicative of a good run and whether you are running "normal" for your machine.

[dmborek]

base composition vs. cluster generation

Could you elaborate how base composition affects cluster generation?

D.

[csc]

Hi D,

You may have heard (or noticed) that there is a bit of base-bias in Solexa/Illumina based sequencing data. This tends to be of the anti-GC variety. One reason for this is library amplification itself (The Broad recently put out a good paper on this). The other is the clonal amplification of library molecules into clusters. One of the reasons you see the anti-GC bias is that high-GC content sequences seem to not amplify well on the flow cell surface. The clusters don't grow sufficiently to be detected when sequencing. Illumina is supposedly working on the amplification step in the clustering process to ensure better GC representation. AT rich sequences tend to generate clusters of larger size. Several years back, Illumina offered a formula to modify the number of amplification cycles in cluster generation (this was a little easier to do on the original cluster station) to account of AT/GC compositions in biased samples in terms of getting "normal" cluster sizes.