CNVs of the 39 samples were determined using the Affymetrix SNP 6.0 platform (Affymetrix, Santa Clara, CA, USA), which interrogates 1 800 000 copy number probes distributed evenly along the human genome (mean interval: 1.8 kilobases) [55 (link)]. The 1 800 000 probes were annotated with home-made tools querying the Ensembl 77th version with Human Genome HG37 MYSQL database available at
ftp.ensembl.org/pub/release-77/mysql/ . Total copy numbers were computed using the Bioconductor crlmm package [56 ]. A sliding window (length = 11 probes) representing the signal median value was used to smooth artifactual data.
This strategy allowed CNV discovery at high resolution and identification of genes frequently altered in relapsed DLBCLs. Duplications were defined by a copy number ≥2.5, deletions were defined by a copy number ≤1.5 and copy numbers between 1.5 and 2.5 were considered normal, according to an already published method [22 (link)]. We used Java Treeview 3.0 [57 (link)] to visualize the results at a genomic scale.
This strategy allowed CNV discovery at high resolution and identification of genes frequently altered in relapsed DLBCLs. Duplications were defined by a copy number ≥2.5, deletions were defined by a copy number ≤1.5 and copy numbers between 1.5 and 2.5 were considered normal, according to an already published method [22 (link)]. We used Java Treeview 3.0 [57 (link)] to visualize the results at a genomic scale.