How to create matrix for leave-one-out-cross validation?

I have dataset that turned into 3500x400 matrix. I want to apply leave one out cross validation. But i cant seem to generate the right train-matrix. Please help.

    mix=randperm(3500);
for i=1:10% divide into 10sets of 350 rows each
    test = mix(1+350*(i-1):350*i);
    train = mix([1+350*(i-2):350*(i+1)-1 350*(i+1):3500]);
    train_labels = Labels(train);
    test_labels = Labels(test);
    %train model
    nb(i) = fitcnb(traina, train_labels,'DistributionNames','mn');
    y(i) = predict(nb,test);
    %Check if prediction was correct
    accuracy(i) = test_labels/350;

end