MiSeq still underclustered at 18 pM
Hello dear people,
We are doing some whole genome sequencing on a plant genome using the MiSeq with v3 reagents for 2x250 paired-end. Library size was measured using a Bioanalyzer and concentration by Qubit.
On the first run, we loaded 15pM library (Nextera), spiked with 1% phiX. We got nice quality (88% >Q30), but a fairly low clustering density (639k/mm2), resulting in 7.3 Gbp of non-index yield.
Surprisingly, the metrics showed 2.3% ‘aligned’ reads, though only 1% should be phiX.
Our second MiSeq run used the same library as before. We tried 18pM but went down to 0.5% phiX. The clustering density has increased a bit, but is still on the low side (738k/mm2), quality has not dropped (it’s about 91% >Q30), and we got 8.26 Gbp non-index reads.
This makes me wonder whether we are somehow overestimating the effective library concentration due to either underestimating average fragment length (I had quite a broad distribution) or incomplete adapter ligation, causing some fragments to wash off without binding to the flowcell.
In the second run, ‘aligned’ reads are still 1.3% despite using 0.5% phiX. Does that mean the effective library concentration that is being sequenced isn’t actually 18pM, but more like 6.9 pM (because 0.5% of 18 pM = 1.3% of 6.9pM) and we should go with a significantly higher concentration?
If with v3 reagents we are aiming for a cluster density of 1200-1400, and 12-15 Gbp of yield, are we seriously underclustering and could get much more out of one run by nearly doubling the library concentration? Is there any downside to loading (supposedly) 30pM if more than half of it washes off the flowcell anyway?
Would love to hear what you think.
We are doing some whole genome sequencing on a plant genome using the MiSeq with v3 reagents for 2x250 paired-end. Library size was measured using a Bioanalyzer and concentration by Qubit.
On the first run, we loaded 15pM library (Nextera), spiked with 1% phiX. We got nice quality (88% >Q30), but a fairly low clustering density (639k/mm2), resulting in 7.3 Gbp of non-index yield.
Surprisingly, the metrics showed 2.3% ‘aligned’ reads, though only 1% should be phiX.
Our second MiSeq run used the same library as before. We tried 18pM but went down to 0.5% phiX. The clustering density has increased a bit, but is still on the low side (738k/mm2), quality has not dropped (it’s about 91% >Q30), and we got 8.26 Gbp non-index reads.
This makes me wonder whether we are somehow overestimating the effective library concentration due to either underestimating average fragment length (I had quite a broad distribution) or incomplete adapter ligation, causing some fragments to wash off without binding to the flowcell.
In the second run, ‘aligned’ reads are still 1.3% despite using 0.5% phiX. Does that mean the effective library concentration that is being sequenced isn’t actually 18pM, but more like 6.9 pM (because 0.5% of 18 pM = 1.3% of 6.9pM) and we should go with a significantly higher concentration?
If with v3 reagents we are aiming for a cluster density of 1200-1400, and 12-15 Gbp of yield, are we seriously underclustering and could get much more out of one run by nearly doubling the library concentration? Is there any downside to loading (supposedly) 30pM if more than half of it washes off the flowcell anyway?
Would love to hear what you think.
Comment