Different reference and variant allele in IonTorrent variantCaller

Question

[ceruleanivy]

I am doing a targeted amplicon NGS analysis on cholangiocarcinoma FFPE samples on proton sequencer on a run with 88% ISP loading. The XLS file provided, containing all the barcodes that went into analysis with the variantCaller (v5.0.2.1) software in IonServer has two different columns for reference and variant alleles that seem to differ in a portion of cases within the same row. For my example the differences were observed in 50/1743 variants. How should I treat these variants ? Keep one of the pair of values or reject them altogether ?

Example

[Titus]

Hi, Strange results. Did you check on IGV for example if there are the 2 variants on the position ?

Best

Answer 1

[geocarvalho]

I think it is a problem related to left and right-normalization to represent indels in VCF. I recommend read this article from Annovar web-site.

[ceruleanivy]

Thanks for your response. Although I use ionReporter for annotation, I recently decided to switch to annovar and I have to admit that I've never looked up the VCF Processing Guide on the website. From my understanding of it, I should perhaps drop the VCF columns and also consider removing some variants completely like the one in row 4. What do you think ?