Re: is there a way to change fit to csv

Hi, In case one wishes to read a NONMEM table containing multiple table headers (e.g. as created using $SIMULATION with more than one subproblem) then you need to remove the unwanted table headers e.g. read.nm<-function(fileName) { dat=read.csv(fileName,sep="",header=T,skip=1,stringsAsFactors=F) #remove TABLE and data item header from subsequent sub-problems temp <- dat[dat[,1]!="TABLE"&dat[,1]!=names(dat[1]),] temp <- as.data.frame(apply(temp,2,as.numeric)) return(temp) } d<-read.nm("table.tbl") > Chandramouli, > > Justin's solution is elegant and works on most datasets produced. However, occasionally NONMEM outputs "numbers" formatted like the following 1.00-103 which should be 1.00E+103. I have created a small function that reads in NONMEM datasets and fixes this problem: > > read.nm<-function(file,na.strings=c("NA","."),...){ > > nmf<-readLines(file,n=-1)[-1] > > nmf<-gsub("+",",",nmf) > > nmf<-gsub("^,","",nmf) > > nmf<-gsub("([0-9])([-+])([0-9])","\\1E\\2\\3",nmf) > > tf<-tempfile() > > cat(nmf,file=tf,sep="\n") > > d<-read.csv(tf,na.strings=na.strings,...) > > unlink(tf) > > return(d) > > } > > Once this function is sourced, all you need to do is read NONMEM in through the new function: > > d<- read.nm("table.tbl") > > Matt. > > *From:* [email protected] [ mailto: [email protected] ] *On Behalf Of *Justin Wilkins > > *Sent:* Tuesday, September 14, 2010 1:24 AM > *To:* Chandramouli Radhakrishnan > *Cc:* [email protected] > *Subject:* Re: [NMusers] is there a way to change fit to csv > > Dear Chandramouli > > Typically something like > > > fit <- read.table("run1.fit", head=T, skip=1) > > (where run1.fit is a typical NONMEM table file, with headers) should do what you need. If not, have a look at the NONMEM table in a text editor - NONMEM tables are basically space-delimited text files, which R's read.table function ought to be able to handle easily. > > Best > Justin >