EDIT: I am updating this question to make it more specific to my issue.

For context - original question prior to edit:

How do I obtain a deamination metric when doing the variant calling using the IonTorrent variant caller, and secondly, how do I correct my called variants for deamination to ensure that these don't provide false positives for the downstream analysis?


I'm calling variants using the IonTorrent TorrentSuite on DNA which has been sequenced from formalin-fixed paraffin-embedded (FFPE) tissue. This has a major issue in that without addition of uracil-N-glycosylase, some of the Ts in the original DNA are deaminated to Cs, which upon sequencing and calling variants can show up as mutations, either as T>C transitions or G>A (from the opposite strand, due to PCR in the library prep). I do not have any idea how long these samples were stored without UNG before sequencing.

TVC (Torrent Variant Caller) gives a deamination metric (essentially, sum of T>C and G>A variants over all variants called), and for our samples, the highest value seen is ~0.92. Naively postprocessing the variants show that for these samples, C>T/T>C transitions overwhelm the remaining variants among my samples.


My use case is this: these are medical samples, which have been inspected by a pathologist (hence the FFPE treatment), and I want to determine which variants are predictive of outcome, hence I have two potentially contradictory goals: reduce false positives and capture the rarer variants which may hold predictive power.

T>C and C>T transitions


  • Given the IonTorrent variant calling pipeline (sequencing > BAM file > TVC > VCF file with deamination statistic):
    • (Q1) Is there a way of correcting the output VCF to remove transition and deamination errors?
    • (Q2) Alternatively, is there a set of filters to use in bcftools to reduce the effect of the errors on how variants are called?
  • $\begingroup$ what is your intended downstream use of the reads? It may not be necessary to correct them at all. $\endgroup$ – conchoecia Jan 30 '19 at 21:12
  • $\begingroup$ I want to take my samples and make inferences about which variants correlate with differing responses to treatment in a more clinical setting, so not having any signal swamped by this C>T T>C noise is what I require $\endgroup$ – user36196 Feb 1 '19 at 13:51

Sites that were deaminated are often called abasic sites.

Q1: How to obtain DNA damage metrics?

You might want to check out mapDamage - this software is used to perform these sorts of analyses and produce plots of types of DNA damage along sequencing reads.

Q2: How to correct for abasic sites and other DNA damage errors?

You can use your choice of kmer-based or alignment based error correction software. The errors will be random and if the sequencing depth is sufficient the erroneous kmers at the ends of reads will be corrected. However, depending on what you are trying to do downstream with your reads you may be better served by not correcting the reads, then dealing with transitions/damage later.

| improve this answer | |
  • $\begingroup$ That's fine but I'm trying to combine some older IonTorrent called variants with variants I've had to call myself, so I don't have access to the original mapped read files for all the samples. I was essentially asking for which settings to try to get the deamination estimate, and how to correct for it once the vcf's have been produced. I want to take my c. 151 samples and make inferences about which variants correlate with differing responses to treatment in a more clinical setting, so not having any signal swamped by this C>T T>C noise is what I require $\endgroup$ – user36196 Jan 31 '19 at 11:09
  • $\begingroup$ I have since found the deamination statistics as provided by TVC (the TorrentSuite variant caller) but still would like to know how to deal with the number that is given and how to filter my VCFs to ignore the false positives occurring due to this mechanism. $\endgroup$ – user36196 Feb 1 '19 at 15:30

Having inspected one of my samples (with estimated deamination metric, the proportion of called SNPs consisting of C>T or G>A transitions, of 0.837), this looks like a straightforward filtering procedure. Looking at the minor allele fraction (AF) and the quality score (-10 log10(p-value)) per type of SNP:AF v QUAL per SNP type it can be seen that the C>T and G>A transitions likely caused by deamination (and PCRing up the deaminated DNA to get the libraries) all cluster in the bottom right of the plot - low AF and low QUAL, so setting thresholds to the left (higher QUAL) and above (higher AF) the groups will remove these from the called variants.

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.