Acceptable Passing Filter (%PF)

[dross11]

I am running some bisulfite converted amplicon-seq libraries (Fluidigm Access Array) on a MiSeq with version2 300-cycle reagents without PhiX (I will be spiking in 5% PhiX in future runs). As this is a low diversity pool and it has been bisulfite treated, I have been expecting lower than normal quality scores, however, I am seeing a rather low %PF. I have run two pools so far which have the following run metrics:

cluster density %PF AVG%Q30
Run1 935K/mm2 75.93% 88.84%
Run2 695K/mm2 65.4% 85.36%

I am not sure whether I should be concerned about the low %PF or whether it is expected for bisulfite converted amplicon libraries. Is there anything I can do to help increase the %PF? Can I still continue with analysing the resulting FASTQs from these runs or is that a bad idea?

Any help would be greatly appreciated.


Edit:
I checked the optic images and they not overclustered.

The first 22 bp of my insert is the Fluidigm sequencing primer and because the %PF is determined after the first 25 cycles the machine might 'think' all the libraries/clusters are homogenous fooling it into determining the run as being overclustered and giving a low %PF. Does this reasoning seem like an explanation?

Edit2:
Yield of the first run was 5.28Gbp and 4.07Gbp and Illumina advertise the maximum v2 yield as 5.1Gb, suggesting most clusters are in actuality passing filter.

[thermophile]

spiking in phiX should increase your pf% but those numbers aren't horrible given you didn't spike in. For my amplicon runs, I spike in 15% phiX and usually see ~80-90% pf

No reason not to use this data, what's passed filter and been demultiplexed should be fine

[WhatsOEver]

Do you need the fluidigm seq primer for anything in the downstream processing? If not, I would suggest to run 10-20 dark cycles (depends if the primer seq is really at the start of every fragment) in the sequencing protocol. If you have a normal seq complexity after the primer, you will not run into any problems in terms of cluster identification.

[dross11]

I performed a run today. I spiked in 5% PhiX and the %PF is at 73.49% and quality is at ~94% with a cluster density of ~1100K/mm2. I'll try 15% next time but I am fairly certain the lowish %PF is due to the sequencing primers presence within my insert. I will also look into performing 15 dark cycles next time round too (this is really very clever).

Thank you for your help and reassurance