regd. num.mark and seg.mean

[anwar]

Hi,

I understand that num.mark would be number of markers and seg.mean
would be related to CNV of segments.

I am having Complete Genomics data which basically is sequencing CNV.
For seg.mean column I am using CALLED_PLOIDY and substracting with 2
for diploids and with 1 for haploids. I am not sure if this is
correct.

Also what should I use for num.mark? and how would it effect the
visual?

Finally, how is CNV Segment calculated?

Thanks in advance.

Anwar

[Jim Robinson]

Hi,

I would suggest transforming CALLED_PLOIDY as follows for display

seg.mean = log2 ( CALLED_PLOIDY  / EXPECTED_PLOIDY),   where EXPECTED_PLOIDY  is 1 or 2.  

For diploids you don't have to do this at all,  you can just leave CALLED_PLOIDY as is.   If there are no negative values for this column in the entire file IGV assumes it is "raw" copy number and will apply this transform.

The num.mark value does not affect the display,  it is only reported in the popup.    Its an optional column,  you could put any integer value you want there or just leave it out.   The first 4 columns + one signal column, which is the last column,  are all that is required.

Jim

[anwar]

Thank you Jim.

There are no negative values in CALLED_PLODY column. However there is
a separate column CALLED_CNV_TYPE which tells if the record is
amplified (+) or reduced (-) or same (=) based on comparison with
normal ploidy. That means for diploids if the CALLED_PLOIDY is 3 then
the CALLED_CNV_TYPE would represented as '+' and similarly for
diploids if CALLED_PLOIDY is 1, then the CALLED_CNV_TYPE would be
represented as '-'.

As you said that if there are no negative values, IGV transforms the
values. I uploaded data both ways (CALLED_PLOIDY as it is from file
and CALLED_PLOIDY (minus) 2 (for diploid) or 1 (for haploid) ) and saw
that the only different in the visual is in the chrY which basically
is haploid, so when I loaded CALLED_PLOIDY as it is, it is considering
it as deletion as mentioned in CNV Summary. Ideally there should not
be any variation because the CALLED_PLOIDY is 1 for most of the
regions.

Please let me know your thoughts.

Thanks in advance.
Anwar

> > Anwar- Hide quoted text -
>
> - Show quoted text -

[James Robinson]

I've made a note in our issue system to do something about chrY and other non-haploid chromosomes or species, currently you will have to fix this yourself by transforming to a log scale as noted previously or it will interpret it as a deletion. Again this is noted as an issue.

Jim

[anwar]

Thanks a lot Jim.

So, just to understand it clearly regarding the log scale conversion,
I have to do the conversion for diploids as seg.mean = log2
( CALLED_PLOIDY / 2), and haploids as seg.mean = log2
( CALLED_PLOIDY / 1).

Thanks once again for your help.
Regards,
Anwar

On Mar 27, 6:32 pm, James Robinson <jrobi...@broadinstitute.org>
wrote:

[James Robinson]

For diploids you shouldn't have to do anything, it should just work. It will do the log2 scaling when computing color although you can do the calculation if you want to. The only difference should be what you see in popup text. For haploids you will have to do the log2 scaling in advance as it will not scale it correctly, so replace the CALLED_PLOIDY values with log2(called_ploidy).

[anwar]

Got it.

Thank you

Regards,
Anwar

On Mar 27, 10:10 pm, James Robinson <jrobi...@broadinstitute.org>

> >> - Show quoted text -- Hide quoted text -

[anwar]

Hi Jim,

One more thing I wanted to ask you.

If I get a zero for (called_ploidy/expected_ploidy) and if I do log2
transform, I would get -infinity value. Should I consider it as zero?

How does IGV handle such values?

Thanks and Regards,
Anwar

On Mar 27, 10:10 pm, James Robinson <jrobi...@broadinstitute.org>

[Vaishali Chakraborty]

Hi,

I had the same query on how to get the seg.mean values for copy number data. I have data from hg19 so diploid organism and in one of the conversations here it is mentioned for diploid, the called ploid is enough and log scaling is automatically done while loading the data in IGV. However, I can see the exact same values from my seg file showing up in IGV.

Can you please provide some guidance on how to calculate this seg.mean value from the called ploidy data? Also, when the called ploidy is a homozygous deletion and the value is 0, log2 of 0 is infinity, how to handle that?

Looking forward to hearing from you.

Best regards,

Vaishali

[igv-help]

Hi, sorry but this isn't really an IGV question, perhaps "biostars" would be a better forum.