question on "uniCox"
1
0
Entering edit mode
Shi, Tao ▴ 720
@shi-tao-199
Last seen 7.1 years ago
Hi list, I?m testing out uniCox R package (version 1.0, on R2.12.0, WinXP). When I ran uniCox on my data, there are always some NA?s in the beta matrix, which in turn causes problems in uniCoxCV call. I don?t see anything wrong with the corresponding data (e.g. no NAs) and if I fit a univariate Cox model, the features that give NA beta estimates are actually pretty significant. Could you please let me know what happened and how to avoid this? I?ve attached the outputs of the function calls below. Thank you very much! ...Tao > a <- uniCox(x=t(dat.ave.train.base), y=sampleinfo.ave.train.base$tm2dthr, >status=sampleinfo.ave.train.base$censrdth) lambda value 1 out of 20 lambda value 2 out of 20 lambda value 3 out of 20 lambda value 4 out of 20 lambda value 5 out of 20 lambda value 6 out of 20 lambda value 7 out of 20 lambda value 8 out of 20 lambda value 9 out of 20 lambda value 10 out of 20 lambda value 11 out of 20 lambda value 12 out of 20 lambda value 13 out of 20 lambda value 14 out of 20 lambda value 15 out of 20 lambda value 16 out of 20 lambda value 17 out of 20 lambda value 18 out of 20 lambda value 19 out of 20 lambda value 20 out of 20 5 betas missing > aa <- uniCoxCV(a, x=t(dat.ave.train.base), >y=sampleinfo.ave.train.base$tm2dthr, status=sampleinfo.ave.train.base$censrdth) FOLD= 1 lambda value 1 out of 20 lambda value 2 out of 20 lambda value 3 out of 20 lambda value 4 out of 20 lambda value 5 out of 20 lambda value 6 out of 20 lambda value 7 out of 20 lambda value 8 out of 20 lambda value 9 out of 20 lambda value 10 out of 20 lambda value 11 out of 20 lambda value 12 out of 20 lambda value 13 out of 20 lambda value 14 out of 20 lambda value 15 out of 20 lambda value 16 out of 20 lambda value 17 out of 20 lambda value 18 out of 20 lambda value 19 out of 20 lambda value 20 out of 20 3 betas missing 1 Error in coxph(Surv(y[ii], status[ii]) ~ eta.new) : No (non-missing) observations > a[[2]][rowSumsis.na(a[[2]])))>0,] [,1] [,2] [,3] [,4] [,5] [,6] [,7] [,8] [,9] [,10] [,11] [,12] [,13] [,14] [,15] [,16] [,17] [,18] [,19] [,20] [1,] 92.6641 NaN NaN NaN 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 [2,] NaN 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 [3,] 567.3650 NaN 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
• 905 views
0
Entering edit mode
@vincent-j-carey-jr-4
Last seen 7 weeks ago
United States
uniCox is not a bioconductor package. the question would probably do better on r-help, or you could contact the uniCox maintainer. On Sun, Nov 21, 2010 at 9:51 AM, Shi, Tao <shidaxia at="" yahoo.com=""> wrote: > Hi list, > > I?m testing out uniCox R package (version 1.0, on R2.12.0, WinXP). > > When I ran uniCox on my data, there are always some NA?s in the beta matrix, > which in turn causes problems in uniCoxCV call. ?I don?t see anything ?wrong > with the corresponding data (e.g. no NAs) and if I fit a ?univariate Cox model, > the features that give NA beta estimates are ?actually pretty significant. > Could you please let me know what ?happened and how to avoid this? > > I?ve attached the outputs of the function calls below. > > Thank you very much! > > > ...Tao > > >> a <- uniCox(x=t(dat.ave.train.base), y=sampleinfo.ave.train.base$tm2dthr, >>status=sampleinfo.ave.train.base$censrdth) > lambda value ?1 out of ?20 > lambda value ?2 out of ?20 > lambda value ?3 out of ?20 > lambda value ?4 out of ?20 > lambda value ?5 out of ?20 > lambda value ?6 out of ?20 > lambda value ?7 out of ?20 > lambda value ?8 out of ?20 > lambda value ?9 out of ?20 > lambda value ?10 out of ?20 > lambda value ?11 out of ?20 > lambda value ?12 out of ?20 > lambda value ?13 out of ?20 > lambda value ?14 out of ?20 > lambda value ?15 out of ?20 > lambda value ?16 out of ?20 > lambda value ?17 out of ?20 > lambda value ?18 out of ?20 > lambda value ?19 out of ?20 > lambda value ?20 out of ?20 > 5 ?betas missing > >> ?aa <- uniCoxCV(a, x=t(dat.ave.train.base), >>y=sampleinfo.ave.train.base$tm2dthr, ?status=sampleinfo.ave.train.base$censrdth) > FOLD= 1 > lambda value ?1 out of ?20 > lambda value ?2 out of ?20 > lambda value ?3 out of ?20 > lambda value ?4 out of ?20 > lambda value ?5 out of ?20 > lambda value ?6 out of ?20 > lambda value ?7 out of ?20 > lambda value ?8 out of ?20 > lambda value ?9 out of ?20 > lambda value ?10 out of ?20 > lambda value ?11 out of ?20 > lambda value ?12 out of ?20 > lambda value ?13 out of ?20 > lambda value ?14 out of ?20 > lambda value ?15 out of ?20 > lambda value ?16 out of ?20 > lambda value ?17 out of ?20 > lambda value ?18 out of ?20 > lambda value ?19 out of ?20 > lambda value ?20 out of ?20 > 3 ?betas missing > 1 > Error in coxph(Surv(y[ii], status[ii]) ~ eta.new) : > ?No (non-missing) observations > >> a[[2]][(rowSumsis.na(a[[2]])))>0,] > ? ? ? ? [,1] [,2] [,3] [,4] [,5] [,6] [,7] [,8] [,9] [,10] [,11] [,12] [,13] > [,14] [,15] [,16] [,17] [,18] [,19] [,20] > [1,] ?92.6641 ?NaN ?NaN ?NaN ? ?0 ? ?0 ? ?0 ? ?0 ? ?0 ? ? 0 ? ? 0 ? ? 0 ? ? 0 > 0 ? ? 0 ? ? 0 ? ? 0 ? ? 0 ? ? 0 ? ? 0 > [2,] ? ? ?NaN ? ?0 ? ?0 ? ?0 ? ?0 ? ?0 ? ?0 ? ?0 ? ?0 ? ? 0 ? ? 0 ? ? 0 ? ? 0 > 0 ? ? 0 ? ? 0 ? ? 0 ? ? 0 ? ? 0 ? ? 0 > [3,] 567.3650 ?NaN ? ?0 ? ?0 ? ?0 ? ?0 ? ?0 ? ?0 ? ?0 ? ? 0 ? ? 0 ? ? 0 ? ? 0 > 0 ? ? 0 ? ? 0 ? ? 0 ? ? 0 ? ? 0 ? ? 0 > > > > > _______________________________________________ > Bioconductor mailing list > Bioconductor at stat.math.ethz.ch > https://stat.ethz.ch/mailman/listinfo/bioconductor > Search the archives: http://news.gmane.org/gmane.science.biology.informatics.conductor
0
Entering edit mode
Thanks, Vincent! I'm aware of that, but just thought this is probably more relevant group. I'll try r-help too. ...Tao ----- Original Message ---- > From: Vincent Carey <stvjc at="" channing.harvard.edu=""> > To: "Shi, Tao" <shidaxia at="" yahoo.com=""> > Cc: BioC <bioconductor at="" stat.math.ethz.ch=""> > Sent: Sun, November 21, 2010 1:19:41 AM > Subject: Re: [BioC] question on "uniCox" > > uniCox is not a bioconductor package. the question would probably do > better on r-help, or you could contact the uniCox maintainer. > > On Sun, Nov 21, 2010 at 9:51 AM, Shi, Tao <shidaxia at="" yahoo.com=""> wrote: > > Hi list, > > > > I?m testing out uniCox R package (version 1.0, on R2.12.0, WinXP). > > > > When I ran uniCox on my data, there are always some NA?s in the beta matrix, > > which in turn causes problems in uniCoxCV call. I don?t see anything wrong > > with the corresponding data (e.g. no NAs) and if I fit a univariate Cox >model, > > the features that give NA beta estimates are actually pretty significant. > > Could you please let me know what happened and how to avoid this? > > > > I?ve attached the outputs of the function calls below. > > > > Thank you very much! > > > > > > ...Tao > > > > > >> a <- uniCox(x=t(dat.ave.train.base), y=sampleinfo.ave.train.base$tm2dthr, > >>status=sampleinfo.ave.train.base$censrdth) > > lambda value 1 out of 20 > > lambda value 2 out of 20 > > lambda value 3 out of 20 > > lambda value 4 out of 20 > > lambda value 5 out of 20 > > lambda value 6 out of 20 > > lambda value 7 out of 20 > > lambda value 8 out of 20 > > lambda value 9 out of 20 > > lambda value 10 out of 20 > > lambda value 11 out of 20 > > lambda value 12 out of 20 > > lambda value 13 out of 20 > > lambda value 14 out of 20 > > lambda value 15 out of 20 > > lambda value 16 out of 20 > > lambda value 17 out of 20 > > lambda value 18 out of 20 > > lambda value 19 out of 20 > > lambda value 20 out of 20 > > 5 betas missing > > > >> aa <- uniCoxCV(a, x=t(dat.ave.train.base), > >>y=sampleinfo.ave.train.base$tm2dthr, > status=sampleinfo.ave.train.base$censrdth) > > FOLD= 1 > > lambda value 1 out of 20 > > lambda value 2 out of 20 > > lambda value 3 out of 20 > > lambda value 4 out of 20 > > lambda value 5 out of 20 > > lambda value 6 out of 20 > > lambda value 7 out of 20 > > lambda value 8 out of 20 > > lambda value 9 out of 20 > > lambda value 10 out of 20 > > lambda value 11 out of 20 > > lambda value 12 out of 20 > > lambda value 13 out of 20 > > lambda value 14 out of 20 > > lambda value 15 out of 20 > > lambda value 16 out of 20 > > lambda value 17 out of 20 > > lambda value 18 out of 20 > > lambda value 19 out of 20 > > lambda value 20 out of 20 > > 3 betas missing > > 1 > > Error in coxph(Surv(y[ii], status[ii]) ~ eta.new) : > > No (non-missing) observations > > > >> a[[2]][(rowSumsis.na(a[[2]])))>0,] > > [,1] [,2] [,3] [,4] [,5] [,6] [,7] [,8] [,9] [,10] [,11] [,12] [,13] > > [,14] [,15] [,16] [,17] [,18] [,19] [,20] > > [1,] 92.6641 NaN NaN NaN 0 0 0 0 0 0 0 0 >0 > > 0 0 0 0 0 0 0 > > [2,] NaN 0 0 0 0 0 0 0 0 0 0 0 >0 > > 0 0 0 0 0 0 0 > > [3,] 567.3650 NaN 0 0 0 0 0 0 0 0 0 0 >0 > > 0 0 0 0 0 0 0 > > > > > > > > > > _______________________________________________ > > Bioconductor mailing list > > Bioconductor at stat.math.ethz.ch > > https://stat.ethz.ch/mailman/listinfo/bioconductor > > Search the archives: >http://news.gmane.org/gmane.science.biology.informatics.conductor >