R help - Mar 2010 - G-test : log-likelihood ratio test

Dear Peter,

I am running your g.test() with the william's correction but I have a
question about the input numbers. These are my data:

"Our data are consistent with those obtained using microarray comparative
genome hybridization in that we found significantly fewer variants per Mb on the
X compared to the autosomal chromosomes (152 versus 336 respectively, G = 93.4,
P < 2e-16, df = 1), and in exons (including UTRs) compared to non-exonic
regions (69 versus 392 respectively, G = 14,226, P < 2e-16, df = 1)."

How do I have to write the input? Should I have to include any total number (in
this case Mb) minus the variants?

g.test(rbind(c(152,(999848)),c(336,(999664))), correct = "williams")
or g.test(rbind(c(152,336)), correct = "williams") ?

Thanks a lot for your help,
Carine

--------------------------------------------
Carine Gubelmann, PhD student
EPFL, Institute of Bioengineering
Bldg. AI 1351
Station 15
CH-1015 Lausanne
Phone: + 41 (0) 21 693 09 59
Fax: + 41 (0) 21 693 09 80


	[[alternative HTML version deleted]]

Can you please, do not write my question on the internet server,
Thanks

De : Gubelmann Carine
Envoyé : mardi, 16. mars 2010 13:36
À : 'r-help@stat.math.ethz.ch'
Objet : [R] G-test : log-likelihood ratio test

Dear Peter,

I am running your g.test() with the william's correction but I have a
question about the input numbers. These are my data:

"Our data are consistent with those obtained using microarray comparative
genome hybridization in that we found significantly fewer variants per Mb on the
X compared to the autosomal chromosomes (152 versus 336 respectively, G = 93.4,
P < 2e-16, df = 1), and in exons (including UTRs) compared to non-exonic
regions (69 versus 392 respectively, G = 14,226, P < 2e-16, df = 1)."

How do I have to write the input? Should I have to include any total number (in
this case Mb) minus the variants?

g.test(rbind(c(152,(999848)),c(336,(999664))), correct = "williams")
or g.test(rbind(c(152,336)), correct = "williams") ?

Thanks a lot for your help,
Carine


	[[alternative HTML version deleted]]