search

WGLab / PennCNV

Copy number vaiation detection from SNP arrays
http://penncnv.openbioinformatics.org
Other
89 stars 53 forks source link

LOH detection #11

Closed romanhaa closed 7 years ago

romanhaa commented 8 years ago

It's me again :) I'm curious why the LOH detection is not a feature anymore and if it has ever been for Affymetrix chips. I feel like there is a real lack of software for this kind of analysis.

kaichop commented 8 years ago

You can use PennCNV for LOH analysis. Just need to make a simple change to the HMM file (hhall.hmm), change the 100 to 0.

On Wed, Aug 3, 2016 at 4:09 AM, romanhaa notifications@github.com wrote:

It's me again :) I'm curious why the LOH detection is not a feature anymore and if it has ever been for Affymetrix chips. I feel like there is a real lack of software for this kind of analysis.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/WGLab/PennCNV/issues/11, or mute the thread https://github.com/notifications/unsubscribe-auth/AFptuDh3tBA01QbCF1wPsUFP4KrvrS05ks5qcEy-gaJpZM4JbaI2 .

romanhaa commented 8 years ago

Oh, you mean in B1_mean? And then I run detect_cnv.pl with '--loh' ? Because it says that parameter is obsolete and might be removed in future releases.

EDIT: I ran this now, and not surprisingly I'm getting several states ranging from 0 to 3. How do I interpret these states?

kaichop commented 8 years ago

Yes.

On Thursday, August 4, 2016, romanhaa notifications@github.com wrote:

Oh, you mean in B1_mean? And then I run detect_cnv.pl with '--loh' ? Because it says that parameter is obsolete and might be removed in future releases.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/WGLab/PennCNV/issues/11#issuecomment-237473085, or mute the thread https://github.com/notifications/unsubscribe-auth/AFptuEA6lM7q-TVppcud5q7qpVVsP-J5ks5qcZKUgaJpZM4JbaI2 .

romanhaa commented 8 years ago

In this scenario, what do the states mean? They again range from 0 to 3. Is it the CN at that LOH call or again an internal parameter?

kaichop commented 8 years ago

state4 means LOH. You should see a lot of state4 cn=2 calls.

On Fri, Aug 5, 2016 at 4:07 AM, romanhaa notifications@github.com wrote:

In this scenario, what do the states mean? They again range from 0 to 3. Is it the CN at that LOH call or again an internal parameter?

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/WGLab/PennCNV/issues/11#issuecomment-237783600, or mute the thread https://github.com/notifications/unsubscribe-auth/AFptuLwOWkNkL32hcM0Isjbeu-wYCVBnks5qcu9cgaJpZM4JbaI2 .

romanhaa commented 8 years ago

Hmm, interesting, I only see the state, no column for CN. Will run it again without "-tabout" even though I don't think that should make a difference. Or do I have to see which CN the LOH calls correspond to from the previous CN analysis (without "-LOH")?

kaichop commented 8 years ago

You did not provide any detail in your question so I have no way of knowing exactly what you are talking about. By default -tabout is not added.

On Fri, Aug 5, 2016 at 10:23 AM, romanhaa notifications@github.com wrote:

Hmm, interesting, I only see the state, no column for CN. Will run it again without "-tabout" even though I don't think that should make a difference. Or do I have to see which CN the LOH calls correspond to from the previous CN analysis (without "-LOH")?

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/WGLab/PennCNV/issues/11#issuecomment-237863896, or mute the thread https://github.com/notifications/unsubscribe-auth/AFptuNr8hmeQ9shBu6x5qNIkwzOgp_GBks5qc0dcgaJpZM4JbaI2 .

romanhaa commented 8 years ago

Sorry, I will try again with some more detail. I first did the normal detect_cnv.pl analysis (meaning without adding '-loh'). In the output file, I had a column with 'state' and one with 'CN'.

chr4:94375102-94525333 numsnp=90 length=150,232 state2,cn=1 GSM... startsnp=SNP_A-4290445 endsnp=CN_1099144

However, when running the same script with '-loh -tabout' (and the modified hmm file), the output file only contains one column that could either be 'state' or 'CN'. Here is an example:

8 13780778 13783440 4 GSM... SNP_A-8503123 SNP_A-8349269

When you said that I should see state4 cn=2 calls, I was expecting two columns. Now my question would have been, if the one column I have is the state, which I have to compare manually to the CN in that region. However, I already started another run without '-tabout' and it gives me the expected output:

chr14:21686225-22053063 numsnp=318 length=366,839 state4,cn=2 GSM... startsnp=CN_635983 endsnp=SNP_A-8656339

So everything fine now, just looks to me like the CN is not included in the output when using '-tabout'.

kaichop commented 8 years ago

Okay thanks for the explanation. -tabout is typically not used by most people.

On Fri, Aug 5, 2016 at 10:45 AM, romanhaa notifications@github.com wrote:

Sorry, I will try again with some more detail. I first did the normal detect_cnv.pl analysis (meaning without adding "-loh"). In the output file, I had a column with 'state' and one with 'CN'.

chr4:94375102-94525333 numsnp=90 length=150,232 state2,cn=1 GSM... startsnp=SNP_A-4290445 endsnp=CN_1099144

However, when running the same script with '-loh -tabout' (and the modified hmm file), the output file only contains one column that could either be 'state' or 'CN'. Here is an example:

8 13780778 13783440 4 GSM... SNP_A-8503123 SNP_A-8349269

When you said that I should see state4 cn=2 calls, I was expecting two columns. Now my question would have been, if the one column I have is the state, which I have to compare manually to the CN in that region. However, I already started another run without '-tabout' and it gives me the expected output:

chr14:21686225-22053063 numsnp=318 length=366,839 state4,cn=2 GSM... startsnp=CN_635983 endsnp=SNP_A-8656339

So everything fine now, just looks to me like the CN is not included in the output when using '-tabout'.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/WGLab/PennCNV/issues/11#issuecomment-237869887, or mute the thread https://github.com/notifications/unsubscribe-auth/AFptuJNuKvJfuzf_ZaCNFsU5IjzG7E4Bks5qc0ymgaJpZM4JbaI2 .