Can you pre-process using awk before loading into R? If so, say you want columns 2,3 and 5, you can do:
awk '{print $2,$3,$5}' yourfile.csv > cols23and5.csv
If your CSV file is quoted (e.g. because some fields contain commas), and looks like this:
"Field 1","Field 2, with commas, in it","Field 3","Field 4, also with commas,,,"
"Field 1","Field 2, with commas, in it","Field 3","Field 4, also with commas,,,"
you can remove the double quotes and change the field separator from commas into, say colons, like this:
sed -e 's/","/:/g' -e 's/"//g' yourfile.csv > ColonSeparated.csv
so that your file becomes:
Field 1:Field 2, with commas, in it:Field 3:Field 4, also with commas,,,
Field 1:Field 2, with commas, in it:Field 3:Field 4, also with commas,,,
then you can process it with awk
using the colon as a separator and without the embedded commas worrying you:
awk -F: '{print $2,FS,$3,FS,$4}' ColonSeparated.csv > SmallFileForR.csv