Dear sequenza users,
I am using Sequenza only recently. After the python script, I ran the 3 functions sequenza.extract(), sequenza.fit() and sequenza.results() with default parameters. My data is a set of 5 tumor-normal WES pairs.
Now, I would like to adjust the parameters more precisely. The problem is that I do not clearly understand part of them, even after reading the vignette, the google group and the description of the R functions.
Sorry for the long post, I do not know a lot sequenza and I am not an expert in segmentation.
Here are the parameters:
1) sequenza.extract
2) sequenza.fit
My last concern is the gamma parameter, which seems crucial in segmentation.
Since I have WES data, I chose breaks.method="full" and now I want to determine gamma.pcf and kmin.pcf.
From what I understood, in cancer data, the range varies often from 15 to 40.
I started with 40. From 140 to 40, copy number estimations did not change but 2/5 cellularity estimations changed: from 0.46 to 0.97 and from 0.35 to 0.26.
Do you have a way to determine gamma?
I will try to use the gamma.plot function of copynumber package.
I am surprised to get very distant cellularities from what we believe. We are pretty sure to have high purity, most likely >90%, but sequenza returned range between 0.35 to 0.5 with default settings.
Does someone get close values in its data?
Sorry for the high number of questions!
Thank you in advance for your feedback.
Jane
I am using Sequenza only recently. After the python script, I ran the 3 functions sequenza.extract(), sequenza.fit() and sequenza.results() with default parameters. My data is a set of 5 tumor-normal WES pairs.
Now, I would like to adjust the parameters more precisely. The problem is that I do not clearly understand part of them, even after reading the vignette, the google group and the description of the R functions.
Sorry for the long post, I do not know a lot sequenza and I am not an expert in segmentation.
Here are the parameters:
1) sequenza.extract
- window=10^6
"size of windows used when plotting mean and quartile ranges of depth ratios and B-allele frequencies. Smaller windows will take more time to compute"
Is this parameter useful only for plotting?
I changed it to 500 and I did not see changes when looking at the genome_view.pdf - overlap=1
"integer specifying the number of overlapping windows"
If we consider a specific window, it can overlap only 0, 1 or 2 window(s), right? - min.type.freq=0.9
"minimum frequency of aberrant types"
What does it mean? - weighted.mean=TRUE
"boolean to select if the segments should be calculated using the read depth as weights to calculate depth ratio and B-allele frequency means"
What does this mean?
2) sequenza.fit
- N.ratio.filter=10
"Threshold of minimum number of observation of depth ratio in a segment"
Minimum number of variants in a segment? - N.BAF.filter=1
"threshold of minimum number of observation of B-allele frequency in a segment"
Minimum number of variants in a segment?
Why is the default value not the same as for N.ratio.filter? - segment.filter=3 10^6
"threshold segment length (in base pairs) to filter out short segments, that can cause noise when fitting the cellularity and ploidy parameters. The threshold will not affect the allele-specific segmentation"
Is it the minimum length of a segment?
What is the usual range? 3 10^6 seems big. - ratio.priority=FALSE
"logical, if TRUE only the depth ratio will be used to determine the copy number state, while the Bf value will be used to determine the number of B-alleles"
Does this mean that with FALSE, both depth ratio and BAF will be use to determine copy number and only BAF to determine the number of B-alleles?
My last concern is the gamma parameter, which seems crucial in segmentation.
Since I have WES data, I chose breaks.method="full" and now I want to determine gamma.pcf and kmin.pcf.
From what I understood, in cancer data, the range varies often from 15 to 40.
I started with 40. From 140 to 40, copy number estimations did not change but 2/5 cellularity estimations changed: from 0.46 to 0.97 and from 0.35 to 0.26.
Do you have a way to determine gamma?
I will try to use the gamma.plot function of copynumber package.
I am surprised to get very distant cellularities from what we believe. We are pretty sure to have high purity, most likely >90%, but sequenza returned range between 0.35 to 0.5 with default settings.
Does someone get close values in its data?
Sorry for the high number of questions!
Thank you in advance for your feedback.
Jane
Comment