Ion Proton beta user feedback?

[nhunkapiller]

How are all you beta users out there liking the Ion Proton in terms of workflow convenience compared to the HiSEQ 2500.

What are typical per base sequencing error rates?

How reliable is the platform (emPCR + sequencing) in terms of breakdown rates and run failures? Is this device capable of performing as reliably as the HiSEQ platform in a production lab?

[GenoMax]

We just had two instruments delivered. Once they are installed and working I will report back.

[Mahmoud]

We just had two instruments delivered. Once they are installed and working I will report back.

What is your early feedback now?

[adaptivegenome]

Would love to know how the data compares to Ion Torrent with respect to error/bias.

[GenoMax]

What is your early feedback now?

Following is not very useful but nonetheless:

For various reasons it took some time to get the instruments installed (missing UPS's, other unrelated local facilities issues etc).

We just ran the first set of data on one of the machines. I have not personally analyzed the data but looking through some fastq file that were generated most of the reads appear to be between ~25-75 bp with about 3-5 million reads per sample. This was the very first run on this instrument and thus should be considered with appropriate caution. Generally things improve with additional runs/time.

[steinmann]

Thanks for lettings us know! Was this attempting to do 100 or 200 bp sequencing? Sounds disastrous.

[thomasblomquist]

Following is not very useful but nonetheless:

For various reasons it took some time to get the instruments installed (missing UPS's, other unrelated local facilities issues etc).

We just ran the first set of data on one of the machines. I have not personally analyzed the data but looking through some fastq file that were generated most of the reads appear to be between ~25-75 bp with about 3-5 million reads per sample. This was the very first run on this instrument and thus should be considered with appropriate caution. Generally things improve with additional runs/time.

Eek!

[matth431]

We've tried a couple of the test libraries now and seem to be making progress. Last run gave an average of 89bp read length with 180bp being the maximum. We got 42M reads (about 3.7Gbp). Our issue seems to be loading - only 57% - plus a relatively high proportion of polyclonal ISPs (30%). The latter may be due to losing ISPs during the various washes - we're not sure our centrifuge is pelleting them well enough.

Initial problems were caused by using lab-made NaOH and the fact that our water (even post-purification to 18 meg-ohm) is very soft... we've been advised to spike our wash bottle with 0.1M NaOH as well prior to initialisation. The machine attempts to adjust it to pH 7.7 but we were starting at ~pH5.3! So was running out of NaOH. Pre-spiking brings the base level up to around pH5.8 - 6.0.

[Buzz0r]

Interesting, please keep us up to date on how you progress in the Proton sample prep and the data metrics coming off the machine!

Any idea yet on how the quality looks of the data?

If you sift though articles and post in the net you can find numbers that range pretty far from pathetic (early users, new users) to OMG data (IonTorrent internally). Would be interesting to see what you can achieve!

Cheers Thomas.

[matth431]

Interesting, please keep us up to date on how you progress in the Proton sample prep and the data metrics coming off the machine!

Any idea yet on how the quality looks of the data?

If you sift though articles and post in the net you can find numbers that range pretty far from pathetic (early users, new users) to OMG data (IonTorrent internally). Would be interesting to see what you can achieve!

Cheers Thomas.

This was a bit delayed as the Torrent Server doesn't come pre-loaded with the hg19 reference sequence, but our last control run was similar in output to the one I mentioned above and was aligned using the onboard pipeline: results attached.

Worth noting that the human genome control library supplied by Life Technologies does not seem to be the exact sample used for the hg19 build, so alignment will never be perfect due to variants - there is almost no alignment to the Y chromosome so suspect whoever the library is from is not of the male persuasion...

We're still experiencing "problems" with the auto-pH function - seems to fail the initialisation checks on the dNTPs every time, but this doesn't seem to impact the read quality...

Was hoping we could improve loading to 60-70% but seem to be struggling - tips appreciated - still, I think we're ready to go ahead and try an actual library, so will post back once we have that data.