Skip to main content
. 2017 Nov 10;10:32. doi: 10.1186/s13040-017-0152-6

Fig. 1.

Fig. 1

Generation of controlled datasets. The final dataset of 3200 genes contain 500 genes having altered correlations at different levels. From these, 300 of these genes gradually lose their correlations by adding noise (step 2). Also, 100 genes gradually increase their correlations (step 3 “Gaining Correlations”) while other 100 decrease their correlations (step 3 “Losing Correlations”). In this way, the dataset generated should not contain differential expressed genes