To remove NaN values and to calculate PCA of dataset

Dear sir,

I have a data set of 141x3084 size that includes NaN values terribly. The overall aim is to reduce the dimensions of data set. But after normalize the data set, I can't calculate the score,latent,var. They all are zeros.Why,sir. After that, I calculate PCA without normalization , I get the NaN value as the result though. Seriously!!! How can I solve my problem,sir. It might be easy for you,sir. Thanks all.

[per isakson]

Documentation on PCA says:

By default, pca performs the action specified by the 'Rows','complete' name-value pair 
argument. This option removes the observations with NaN values before calculation. Rows 
of NaNs are reinserted into score and tsquared at the corresponding locations, namely 
rows 56 to 59, 131, and 132.

See also http://uk.mathworks.com/matlabcentral/answers/93137-can-princomp-be-used-on-incomplete-or-missing-data-in-matlab-7-8-r2009a#answer_102485