Simultaneous gene expression profiling by using the same cDNA arrays for all samples showed that 47% of the amplified genes correlate with gene expression (using a correlation coefficient cutoff 0.5; data not shown).
Using a correlation coefficient cutoff of 0.2, six tumors were defined as basal-like, seven were luminal A, three tumors were ERBB2+, one luminal B and finally, one was identified as normal breast-like.
In Stöckel et al. [ 16] diurnal regulated genes were classified using a correlation network and a fold-change cutoff of 1.3, in which genes with a period other than 24 hours were excluded, since they did not correlate with the main oscillatory network.
To address this problem, we filtered the marker pairs by removing all highly correlated pairs from consideration using a Pearson correlation cutoff of r ≥ 0.75.
In order to improve the performance, we developed a hybrid model using all reduced fingerprints at a correlation cutoff value > =0.6 obtained from each class and achieved MCC value 0.28 slightly better than model developed on individual class.
Circadian oscillations were identified using default parameters (see http://haystack.cgrb.oregonstate.edu for details) except that a fold-change (FC) of 1.5 was applied for transcripts, whilst for metabolites a correlation cutoff of 0.7 and FC of 1.2 were used.
Interestingly, a simple method using Pearson correlation cutoff of 0.70, although not as good as meta-analysis, outperformed sophisticated methods of causal linear regression model and graphical Gaussian model.
Using the correlation cutoff (r = 0.59), a total of 598 (89%) paralogous protein families with two-qualifying MPSS genes exhibited divergent expression patterns, consistent with what has been reported in Arabidopsis [ 36] and in yeast in which more than 80% of the older duplicated gene pairs (ds > 1.5) showed divergence in expression [ 41].
An examination of the values indicates that none of the ethnic groups is exactly the same type (using a cutoff intraclass correlation of 0.98 [ 37]), although the Caucasian and Japanese groups are the closest (0.945).
The results presented here are for using a cutoff for the correlation coefficient of 0.4, but qualitatively they are similar for higher cutoffs (only the number of reported pairs is lower).
