To determine the functional consequence of CNAs with regards to gene expression we first selected CNAs for which five (~1%) or more of the cell lines had a copy number different from two (regardless of copy number in the corresponding somatic cells). For each of these CNAs we took the copy number at the region of peak coverage for each cell line (see CNA coverage plots in Fig. 2c and Extended Data Fig. 4). We then defined the set of expression array probes to test by choosing only expressed probes. Here a probe was defined as expressed if the number of RNA-seq fragment counts (normalised between samples for sequencing depth) mapping to the genomic regions targeted by the probe is greater than 0 in 10% or more of the lines. Finally, we used a linear mixed model to independently test for association between copy number of each CNA and the intensity of each probe. We included culture condition, gender and an interaction between copy number and culture condition as fixed effects; and used donor and assay batch as random effects.
