HyPhy message board - How to confirm GARD result?

Sundy

YaBB Newbies

Offline

Feed your monkey!

Posts: 33

Re: How to confirm GARD result?
Reply #15 - Dec 11^th, 2008 at 9:46am

Dear Sergei,
Thank you very much for your interpretation.
Yes, I see. I think I should delete some of the very similar sequence in my datasets, then run GARD. Thanks.

So "it expects the _splits file to start with two blank likes"
Does this means that if there is only one breakpoint, GARDProcessor.bf cannot verify the significant of the breakpoint?

And another question,
After running GARD, the result web-page has a such sentence: "To process RAW files you need to download this HyPhy batchfile, run it locally in HyPhy and input the raw file when prompted."
Could you please tell how to use the "HyPhy batchfile".
Do I need use the "HyPhy batchfile" when I run GARDProcessor.bf?

Thank you!
Best regards,

Sundy

Back to top

IP Logged

Sergei YaBB Administrator Offline Datamonkeys are forever... Posts: 1658 UCSD Gender:	Re: How to confirm GARD result? Reply #16 - Dec 11^th, 2008 at 3:21pm Dear Sundy, You can run GARDProcessor.bf on the output with a single breakpoint: just make sure the _splits file starts with two blanks lines. GARDProcessor.bf is the same as the batchfile that Datamonkey instructs you do download. Cheers, Sergei
Back to top	Associate Professor Division of Infectious Diseases Division of Biomedical Informatics School of Medicine University of California San Diego WWW IP Logged

Sundy YaBB Newbies Offline Feed your monkey! Posts: 33	Re: How to confirm GARD result? Reply #17 - Dec 11^th, 2008 at 6:19pm Dear Sergei, Thank you so much!! I miss understood your meaning and I get the idea now. Best regards, Sundy
Back to top	IP Logged

Sundy YaBB Newbies Offline Feed your monkey! Posts: 33	Re: How to confirm GARD result? Reply #18 - Dec 11^th, 2008 at 6:47pm Dear Sergei, I got another problem (as enclosed jpg), when I run a data as attached file with GARDProcessor.bf. Could you please help me check it? Thank you!! Sundy
Back to top	Multimedia File Viewing and Clickable Links are available for Registered Members only!! You need to Login (58 KB \| ) IP Logged

kevintz

YaBB Newbies

Offline

Feed your monkey!

Posts: 15

Re: How to confirm GARD result?
Reply #19 - Jan 12^th, 2009 at 3:21pm

This site is really helpful. THanks.
I have two question. 18 sequences were analyzed by GARD for recombination breakpoints. 6 breakpoints were found. Then I use the raw file, splits file, and recombinationprocessor for KH testing in HyPhy. When testing the last partition 7, the P-value is 0.1809, while other 6 partitions p-value for KH testing all <0.001, but results showed "A total of 4/6 significant couplings". One partition is gone, 6 left, should it be 5 breakpoints?

From websites, it is said that AICc>10 is significant, but I found breakpoints with an improvement of AIC of 2.7 could be significant, because p-value of either side of sequences for KH testing is less than 0.001. So, I want to know how I can make sure the results? See AICc or just see P-value?

Thanks. I am a fresh on this statistic field. Thanks in advance.

Back to top

IP Logged

Sergei

YaBB Administrator

Offline

Datamonkeys are forever...

Posts: 1658
UCSD
Gender: male

Re: How to confirm GARD result?
Reply #20 - Jan 12^th, 2009 at 3:39pm

Dear kevintz,

A AIC-c improvement of 10 is only a guideline (and probably a rather conservative one), hence it's no surprise that a KH test shows topological discordance for a smaller AIC-c. You can have a major AIC-c improvement but no topological difference due to differing branch lengths (imagine that a part of the alignment evolves 10 times faster than the rest of the alignment - for example a hypervariable region in a gene), hence the need to confirm discordance using the KH test. A strong c-AIC difference coupled with low KH p-values is probably the most reliable indicator that there indeed is recombination in the sample.

As far as your example goes, there should indeed be 5/6 couplings if no other KH test is insignificant. Can you perhaps paste in the entire output of the processor file for me to look at?

Sergei

Cheers,
Sergei

Back to top

Associate Professor
Division of Infectious Diseases
Division of Biomedical Informatics
School of Medicine
University of California San Diego

WWW

IP Logged

kevintz YaBB Newbies Offline Feed your monkey! Posts: 15	Re: How to confirm GARD result? Reply #21 - Jan 13^th, 2009 at 1:42am Thanks for your reply. the following is the file.
Back to top	Multimedia File Viewing and Clickable Links are available for Registered Members only!! You need to Login (26 KB \| ) IP Logged

kevintz

YaBB Newbies

Offline

Feed your monkey!

Posts: 15

Re: How to confirm GARD result?
Reply #22 - Jan 13^th, 2009 at 6:55am

Another question, GARD for breakpoints showed 2 breakpoints with AICc value of 45.53, and 63.1. KH test indicates good evidence of the two breakpoints. I submitted the GARD results directly to datamonkey by press the link below the results, but the datamonkey gave just one partition. What happened? I am confused.

Thanks.

Best regards

Back to top

IP Logged

Sergei YaBB Administrator Offline Datamonkeys are forever... Posts: 1658 UCSD Gender:	Re: How to confirm GARD result? Reply #23 - Jan 13^th, 2009 at 7:34am Dear kevinz, I am afraid I don't understand the question. How did you obtain the two breakpoints originally before submitting them to Datamonkey? Sergei
Back to top	Associate Professor Division of Infectious Diseases Division of Biomedical Informatics School of Medicine University of California San Diego WWW IP Logged

kevintz YaBB Newbies Offline Feed your monkey! Posts: 15	Re: How to confirm GARD result? Reply #24 - Jan 13^th, 2009 at 7:59am Sorry for confusion. I use another dataset for GARD analysis, and get two breakpoints.
Back to top	IP Logged

kevintz YaBB Newbies Offline Feed your monkey! Posts: 15	Re: How to confirm GARD result? Reply #25 - Jan 13^th, 2009 at 7:59am Here is the link.Multimedia File Viewing and Clickable Links are available for Registered Members only!! You need to Login And the KH test results
Back to top	Multimedia File Viewing and Clickable Links are available for Registered Members only!! You need to Login (11 KB \| ) IP Logged

Sergei YaBB Administrator Offline Datamonkeys are forever... Posts: 1658 UCSD Gender:	Re: How to confirm GARD result? Reply #26 - Jan 13^th, 2009 at 8:07am kevintz wrote on Jan 13^th, 2009 at 1:42am: Thanks for your reply. the following is the file. Another p-value (partition 5, 0.0014) becomes greater than the threshold of 0.01 following a Bonferroni correction for multiple tests; hence that partition is judged insignificantly supported. Cheers, Sergei
Back to top	Associate Professor Division of Infectious Diseases Division of Biomedical Informatics School of Medicine University of California San Diego WWW IP Logged

Sergei

YaBB Administrator

Offline

Datamonkeys are forever...

Posts: 1658
UCSD
Gender: male

Re: How to confirm GARD result?
Reply #27 - Jan 13^th, 2009 at 8:10am

kevintz wrote on Jan 13^th, 2009 at 7:59am:

Sorry for confusion. I use another dataset for GARD analysis, and get two breakpoints.

This happened because Datamonkey removed a duplicate sequence from the alignment and the trees provided by GARD now had too many sequences. I meant to add some code for Datamonkey to handle these cases automatically and trim removed species from the tree but never got around to it. Please resubmit the analysis to GARD with the reduced alignment and all should work well.

Cheers,
Sergei

Back to top

Associate Professor
Division of Infectious Diseases
Division of Biomedical Informatics
School of Medicine
University of California San Diego

WWW

IP Logged

kevintz

YaBB Newbies

Offline

Feed your monkey!

Posts: 15

Re: How to confirm GARD result?
Reply #28 - Jan 13^th, 2009 at 8:24am

Sergei wrote on Jan 13^th, 2009 at 8:10am:

kevintz wrote on Jan 13^th, 2009 at 7:59am:

Sorry for confusion. I use another dataset for GARD analysis, and get two breakpoints.

This happened because Datamonkey removed a duplicate sequence from the alignment and the trees provided by GARD now had too many sequences. I meant to add some code for Datamonkey to handle these cases automatically and trim removed species from the tree but never got around to it. Please resubmit the analysis to GARD with the reduced alignment and all should work well.

Cheers,
Sergei

Does that mean I should remove the duplicate sequence, and then use left sequences for GARD analysis and followed datamonkey selection analysis?

Back to top

IP Logged

Sergei YaBB Administrator Offline Datamonkeys are forever... Posts: 1658 UCSD Gender:	Re: How to confirm GARD result? Reply #29 - Jan 13^th, 2009 at 12:50pm Dear kevintz, You're correct. Please filter the duplicate sequences. You can use HyPhy to do it for you (see 2.3.3 of Multimedia File Viewing and Clickable Links are available for Registered Members only!! You need to Login) Cheers, Sergei
Back to top	Associate Professor Division of Infectious Diseases Division of Biomedical Informatics School of Medicine University of California San Diego WWW IP Logged

	Welcome, Guest. Please Login