Dear Dr. Rodney,

We also using Ion Torrent sequencing for research purpose currently, and trying to use the same at clinical diagnosis, with respect to microbial identification and drug resistance.

Ion Torrent is slightly lower end of sequencing quality with respect to other platforms such as Illumina, where the quality and the paired end sequencing will be helpful in mutational calling, where the confidence level is high, which we lack in Iontorrent,

As per my experience, Iontorrent will be good for mutational studies, if the Coverage is above 40-50X with specific to that particular region of our interest (not the whole coverage as 40-100x) 

With this condition we are preparing for ourself for future diagnostics.

Thanks 

REGARDS

Murugan N

Thank you Nandagopal. We feel that the calling has improved greatly with enhancement to the system software.  Also more coverage improves read accuracy.

Do you want to find point mutations or larget-scale structural variations?

We are using it mainly for point mutations however we are looking at using AmpliSeq panels for fusion gene detection. thanks

Our CLIA laboratory uses the Ion PGM and the Cancer Hotspot Panel v2 as a LDT for clinical testing. For that panel, we've found the latest versions of Torrent Suite and variantcaller to be very accurate when you have >50X coverage. Our only issue is with calls made for variants that overlap indels, but that can be corrected in most cases by adjusting either the reference sequences and/or the variant sequences, or by addiing more specific variant definitions, in the bed file used by variantcaller. 

Thank you for your response.  We are also using Hotspot Panel v2 and quite pleased with it thus far.  We have encountered the problematic scenario you describe.  Did you make the adjustments in house or did Life Technologies help out?

In house. You can ask Life Tech for help or look on Ion Community, but we decided to do it ourselves. The changes we made in the bed file are considered proprietary by my company (sorry), but if you know the problematic calls you get and the other hotspots they overlap, in some cases re-adding them as additional enties in the hotspots.bed file after making compensatory extension and/or shortening of the reference sequences can solve some of the issues. Adding some of the overlapping calls that are indicated as "novel" in the output to the bed file will solve other issues. We also apply an algorithm that compares the coverage of all hotspots that overlap defined deletions to make a separate call that they are deleted if the deletion is called. You could also do that as a macro in excel.

Very interesting solution. Do you mind if I share your answer with my bioinformaticians?

That's fine with me.

anyone using pan bacterial panel? with CosmosID?