Representative Spectral Clustering for Large Data Sets applied to Gene Expression Data