[R] question on "uniCox"

[Shi, Tao]

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]][(rowSums(is.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