Why and how: PhiX SPike-in?

[bio_informatics]

Hi Seqanswer team, and forum members,

Pardon me if this seems to be a broad and naive question.

I do not use sequencing kits, neither I do the quality checks of the data that immediately comes out of the sequencer. I deal with the data that comes after all these checks, and thus, I remain oblivious to what goes on behind the scenes before and during sequencing.

I'm seeing PhiX term in 16s rRNA papers, and its concentration.

I found these urls:.

http://support.illumina.com/sequencing/sequencing_kits/phix_control_v3/questions.html

http://res.illumina.com/documents/pr...xcontrolv3.pdf

http://seqanswers.com/forums/showthread.php?t=31419

Concentration varies from sequencing platform, and what you're sequencing. For low diversity sample, sometimes ~10% is suggested sometimes ~50%.

My query:
- I'd like a basic (read as superficial) understanding/definition what is the importance of PhiX?
How does an external element helps to have a better quality data?

- Doesn't it contaminate primers, bar code, indices?

[HESmith]

PhiX serves multiple functions:

1) technical control for clustering reaction (spiking in a known amount of phiX should yield a known number of phiX clusters).
2) technical control for sequencing accuracy (on-the-fly alignment to the phiX reference is used to calculate the sequencing error rate).
3) introduction of sequencing diversity in low-complexity libraries (diversity is needed to discriminate clusters and create signal thresholds for base-calling). As the software has improved, the recommended amount of phiX spike-in has decreased.

The phiX library does not contain an index; these reads are assigned to the Undetermined_indices directory.

[bio_informatics]

Hi HESmith,

Thanks for valuable points and your reply.

Would introducing higher % (~30-50) of PhiX for amplicons, at MiSeq platform would cause bias?

[microgirl123]

Older basecalling software on the MiSeq used to require a 50% phiX spike for low-diversity (amplicon samples). More recently, the software has been updated and only requires a 5-10% spike.

[bio_informatics]

Hi microgirl,
Thanks for your reply.

[GenoMax]

Hi HESmith,

Thanks for valuable points and your reply.

Would introducing higher % (~30-50) of PhiX for amplicons, at MiSeq platform would cause bias?

Why do you think that would happen? phiX does compete for spots on the flowcell (but it should not out-compete specific amplicons) so you do end up losing some sequencing capacity.

[bio_informatics]

Hi Genomax,
Thanks for your reply.

I'm reading posts related to PhiX, joining pieces from here and there, and probably the question is incorrect/naive.

I read:
"Depending on what MiSeq software version was used for the sequencing, an inadequate amount of PhiX combined with too high a cluster density will result in an excessive number of sequencing errors leading to a large number of unique sequences."

I asked wrongly on PhiX concentration, however, it was intended with cluster density.
Again, I do not know much of these terms, and traversing for the sake of my knowledge, and thus muttering these queries.

Why do you think that would happen? phiX does compete for spots on the flowcell (but it should not out-compete specific amplicons) so you do end up losing some sequencing capacity.

[GenoMax]

As @microgirl123 pointed out above there was a period in early MiSeq days when there were problems sequencing low complexity libraries. Most of these issues have been mitigated via software updates for MiSeq Control Software (MCS) over time.

[bio_informatics]

Hi Genomax,

That clears many doubts.
Merci!

[kerryp]

are my phiX QC measures ok?

I've just used phiX (v3) for the first time on a miseq run. I was using the 500-cycle reagent kit (running 241*2 cycles), and using 1% PhiX.

According to my run summary:

aligned = 1.3%
error rate (read 1) = 1.66 (+/- 0.13)
error rate (read 4) = 2.19 (+/- 0.14)

NB for both read 1 and read 2, error rate doubles between cycles 35 - 100

>=Q30 was 83.7% overall

Would these phiX measures fall within acceptable parameters, or should I be concerned? I can't find any clear guidance online as to what they actually mean.

Thanks in advance for any help!

[Jessica_L]

The error rate sounds a little high to me, but that is dependent on other factors, including the cluster density for the run. Can you post the run summary chart and the intensity plot from the data by cycle section (IVC plot)?

[kerryp]

thanks jessica_L
Here's the charts screen - is this what you were looking for?

[GenoMax]

Run looks to be within illumina spec. Have you had a chance to look at the data? Some of the libraries may have short inserts. A typical characteristic manifestation of this is dropping Q-scores towards the end of reads. I assume the run has completed without any problems.

[kerryp]

Haven't looked at the data beyond running it on spades (where there seemed to decently few contigs). But when I ran the library on bioanalyser, the average fragment size for all samples were within the 600-700bp range.
The run finished without any problems.
Glad to hear it may be within illumina spec. Where could I find this information (recommended values/thresholds for QC measures)?
Thanks

[GenoMax]

As you do more runs you will start to develop a feel for what looks good and what does not. There isn't hard and fast thresholds (but a more continuous gradient) as to what constitutes good/bad runs. If you using your MiSeq in your own lab then you will have more control over samples/libraries. This is not possible when the instrument is being used in a core.

In general if you find a significant deviation (in terms of % Pass Filter, Q-scores) beyond the published spec (link below) then you stand a good chance of getting free replacement reagents from Illumina (once Illumina tech support determines that the problem was instrument/reagent related).

You can find published performance specification for MiSeq here: http://www.illumina.com/systems/mise...fications.html