AFAIK, "no can do" with xlsread; it can only pass a single range argument. In Matlab, it's probably simplest to simply import the full sheet then just keep the columns you want (or, conversely, clear the unwanted columns by setting to []).
The syntax for a single column is 'A:A' or '5:5' for column A or row 5, respectively. Unfortunately,
x=xlsread('file.xls','A:A,D:D,F:F')
doesn't error, it just returns column A, ignoring other inputs even though that is syntax which internally inside Excel you could write something like =SUM(A:A,D:D,F:F) in B1, say.
For your specific case,
x=xlsread('file.xls','B:F'); % subgroup
x=x(:,1:2:end); % save columns 1,3,5 which originally were B,D,F
This illustrates some other features of Matlab indexing expressions as well.
In reality, I'd probably write
x=xlsread('file.xls'); % just read the sheet
x= x(:,['B' 'D' 'F']-'@'); % save columns B,D,F
which illustrates yet another feature that you can also do arithmetic on characters. It just so happens I've been around the block enough times that I happen to remember that '@' precedes upper-case 'A' in the ASCII collating sequence w/o having to look it up or compute it. That comes from being far too old, basically... :)
Or, of course you can write if it's simpler indexing
x=xlsread('file.xls'); % just read the sheet
x(:,['A':2:'G']-'@')=[]; % clear columns A,C,E,...
Experiment with the above indexing expressions at the command line to see what they do and how...
