View Single Post
Old 11-29-2016, 05:59 AM   #1
lucorum
Junior Member
 
Location: Moscow, Russia

Join Date: Nov 2016
Posts: 5
Default Poor Ion PI Chip loading - can it possibly be caused by some contamination?

Hello everyone,
I have been working with Ion Torrent Proton sequencer for almost two years now. I do an RNA-Seq pipeline: prepare cDNA libraries (Ion Total RNA-Seq Kit v2), then perform emulsion PCR on One Touch 2 machine, enrich ISPs on One Touch ES machine, then load ISPs on a chip (Ion PI Sequencing 200 Kit v3, Ion PI Chip v2). Every time I do it I perform all these steps identically (at least, I believe so), but some runs (like 20% of all runs) inexplicably fail. Usually it's poor chip loading (less then 20% wells), though I can achieve 80-90% loading in other runs. One or two times loading was mediocre, like 30%, but majority of all the reads were "low quality" reads.
I stick to the protocol quite strictly, the only major deviations are about cleaning chips (it's kinda bad lab tradition). I routinely use cleaning chips for initialization (but only the chips designed for MQ water cleaning, not chlorite cleaning chips); the protocol says explicitly that you shouldn't do it, but doesn't elaborate why. I also sometimes use the same MQ cleaning chip and the same chlorite cleaning chip for months while it's said you should replace them in a week (it's because we usually run the sequencer once a month or even more rarely and I just don't have a used chip from the previous week). However, I perform the full chlorite cleaning almost every time before initialization (this, too, due to the rare use of sequencer), and I've never seen any initialization errors like pH out of range.

I deduced only three factors, that I'm aware of before I start the sequencing run, that can spoil chip loading density. First, it's foam, if the bubbles are too big and you inject too much air, you will lose some ISPs. Second, if despite all precautions there was some electrostatic discharge while picking up the chip, the loading will worsen and you will see distinct rectangular patterns on the loading heatmap. Third, once my libraries were too long (peak lenght 300+ bp instead of 200+), and I believe that either ISPs didn't fit into wells or elongation time wasn't enough during emulsion PCR. Still, either of these factors don't reduce the loading any worse then to 60%, and occasional 10% disaster can't be explained by them.

In my sequence runs, Test fragments Percent 50AQ17 usually is strangely increased in bad runs: it's 80-90% in bad runs and only about 50-60% in good runs (it's supposed to be >90% if everything's OK, isn't it?).

Is it possible that abusing of cleaning chips may result in some contamination and somehow lead to poor loading? Or there are some other factors I should take into account? I've read lots of troubleshooting items and haven't found an answer.

I attach three reports, examples of "good" and "bad" runs here.

If anyone has some bad reports and you know for sure what exactly caused the fail, please, show them to me too.
Attached Files
File Type: pdf good run.pdf (652.2 KB, 23 views)
File Type: pdf the worst run.pdf (524.9 KB, 21 views)
File Type: pdf bad run with low quality.pdf (669.0 KB, 15 views)

Last edited by lucorum; 11-29-2016 at 06:24 AM.
lucorum is offline   Reply With Quote