I'd be very cautious applying these shrinkage methods just because a covariance matrix is "uncooperative" and you get more significant effects.
The {corpcor} methods are meant for small n, large p problems, where covariance estimation is not possible or unstable otherwise.
