Help with the CNVkit analysis
Hi All,
I am confused with the results of CNVkit tool. I am using a flat_reference.cnn against my test sample that has ploidy 4. I am expecting cn = 4 for most of the segments in the .cns file. However, I am getting cn = 2 for most of the segments. Please if someone can explain, why this is the case. Or do I need to change any parameter if the ploidy of the test sample is different from 2? Your quick reply will be much appreciated.
Many thanks, Rav
• 3,475 views
•
link
written
by
kaurravneet4123 •
1 answer
Dear German,
Many thanks for your reply. Do you mean, I should divide the cn values by 4?
Kind regards,
Rav
• 0 views
•
link
written
by
kaurravneet4123 •
Log in to answer this question.
More posts like this
-
Statistical comparison of chr8 amplifications in matched primary vs metastatic WGS samples using lo…
written by Dalal •
-
Using CNVkit for checking normal samples
written by DGTool •
-
how to run multiple bam files for CNVkit input
written by Dora •
-
The exonix length calculated from the CNVkit bed files is several fold larger than the real exonic …
written by ruhollah •
-
how should I filter call.cns results according to cn depth p_ttest probes weight
written by linouhao •
-
Analysis using CNVkit tool
written by kaurravneet4123 •
-
Meaning of mfold upper limit?
written by mbk0asis
-
cnvkit output understanding
written by Folder40g
-
I need help with how to modify the sbdry parameter in DNAcopy
written by mikki •
-
CLC Bio Variant detection with a Polyploid
written by doconnor1874 •
Ploidy of 2x is assumed by default, so yes, Ploidy of 4 has to be specified or you can divide your final results by corresponding factor.
Dear German,
Many thanks for your reply. Do you mean, I should divide the cn values by 4?
Kind regards,
Rav
Hi,
I think you need to find difference from 4 and then divide by 2 and then plus 2 =)
E.g. CN=6 with ploidy 4
(6 - 4) / 2 + 2 = 3 with ploidy 2. That's what is expected.
Hi German,
Many thanks for your answer.
So my understanding is as below :
if cn=6 while using a flat_reference.cnn and a test sample with ploidy = 4, then
6 - 4 = 2 , this would give you the gene gain or loss for a particular segment of a tetraploid sample.
(6-4)/2, this would give gene gain or loss per diploid genome
then why are we adding 2 in the end?
And is my understanding correct for all the steps?
Sure, the understanding is correct! In the beginning we substracted a baseline - 4 copies. So in the end we have to add baseline - 2, as we assume that the organism is diploid. You may skip adding 2 step - but then you'll have a relatvie change and e.g. for X/Y chromosomes it may be crucial.
I understand now. Many thanks for all the clarification.
As you mentioned earlier that ploidy of 4 can be specified in the pipeline also. Where can I specify the ploidy in the pipeline so that I do not need to do the calculations in the end?
Sadly the option I was thinking about is for different reasons. You need to perform calculations yourself, ie.
https://cnvkit.readthedocs.io/en/stable/pipeline.html
Thank you German for all the above clarifications. Do you think it is appropriate to calculate CNV using a tumour cell line with ploidy=4 against a normal cell line with ploidy=2? If yes, then how can we interpret the copy number from the log2 value?
Basically you have no other choice - you don't have a normal with ploidy of 4 =) copy-numbers have to be corrected for ploidy with the above mentioned procedure - change the baseline from 2 to 4, then calculate differences
Changing the baseline fro 2 to 4 has worked. Thank you so much for all your time and help.
Hi German,
While using a tumour cell line with ploidy=4 against a normal cell line with ploidy=2, if the log2 value of a segment is -0.37681, should I interpret it's copy number as 0.77 or 0.38?
Because log2 = -0.37681 and its antilog would be 0.77.
How should I interpret it's copy number? Sorry, I am bit confused.
hi, log2(CN_tumor / CN_normal) = -0.37681. By default CN_normal = 2. CN_tumor then is equal to 1.549604. And then you need to apply that trick as was discussed before - 4 + (1.55 - 2) * 2 = 3.1. Of course I do not know how CNV-Kit does ploidy correction and these calculations are correct only if CN_tumor is assumed to be 2 by default.
Many thanks for your reply.
I was think the following:
if log2(CN_tumor / CN_normal) = -0.37681
and CN_normal = 2,
then, CN_tumor= (2^-0.37681)*2 = 0.770138593 *2 = 1.540277186
Therefore, CN/diploid = 1.540277186/2 = 0.770138593
Now, if CN_normal = 2, then the CN_tumor/diploid = 0.77
Therefore, if CN_normal = 4, then CN_tumor/diploid should be = 0.77/2 = 0.38
I am more inclined towards CN_tumor/diploid = 0.38 because, if I run the analysis of the same tumor sample with ploidy = 4 against a flat reference, then I get the following result for the same segment:
log2 = -0.387476
Antilog = 0.7645
therefore, CN_tumor/diploid = 0.38225
but CN_normal is not 4. it is 2. your normal sample is diploid. however, I do not deeply understand the design and assumptions of calling, so it is up to you to decide! =)