Running the same sample multiple times results in different PASS/FAIL result

[kmavrommatis]

Hi,
I have a very fragmented tumor sample which has been segmented by various methods (FACETS, SCLUST, Battenberg) and i am using CNAqc to check its quality and potentially identify the best approach to analyze this genome.
My thoughts are that given the current segmentation I could use CNAqc to identify the segments that are more likely to be correct vs more likely to be wrong. Furthermore, I am hoping that if a sample fails I can get rerun it with different arguments through the same algorithms e.g. provide a different ploidy estimate etc.

I have noticed though that if I run the same cnaqc object through the analyze_peaks function multiple times I get different results.
For example

Running:

xlist=lapply( 1:10, function(y){analyze_peaks(x, n_bootstrap=10)}) %>% magrittr::set_names( paste0( "rep_",1:10))

I get

sapply(xlist, function(x){
    x$peaks_analysis$QC
})
  rep_1 rep_2 rep_3 rep_4 rep_5 rep_6 rep_7 rep_8 rep_9 rep_10
  <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr> 
FAIL  PASS  FAIL  FAIL  PASS  FAIL  PASS  PASS  FAIL  PASS 

While

xlist2=lapply( 1:10, function(y){analyze_peaks(x)}) %>% magrittr::set_names( paste0( "rep_",1:10))

I get

sapply(xlist2, function(x){
    x$peaks_analysis$QC
})
  rep_1 rep_2 rep_3 rep_4 rep_5 rep_6 rep_7 rep_8 rep_9 rep_10
  <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr> 
PASS  PASS  PASS  PASS  PASS  FAIL  PASS  PASS  PASS  PASS

These samples fail because their purity correction is high (> 8%). However, in the instances where the samples PASS the purity correction is <2%.
Any advice on this?
Thanks
K

[caravagn]

Hi, can you post the basic plots of the genome? segmentation, statistics for the QCed segments, VAFs, peak analysis etc.

It's hard otherwise give you a helpful answer.

G

[caravagn]

@vvvirgy please provide some support if possible.