Why copy number variations of TCGA is so high

Question

[akij]

I was reading some papers on copy number variations. Some of the papers referred to TCGA website from where they downloaded the copy number variation(CNV) data. Most papers mention that on average less than 100 CNVs are found in a cancer cell but when I look at the CNV data of TCGA website almost all of them have more than 100 CNVs in each sample. So why TCGA has such high CNVs in each sample.

[poisonAlien]

how did you look at TCGA cnv data? Was it from GISTIC output ?

[akij]

Sample data

Sample  Chromosome  Start   End Num_Probes  Segment_Mean
AMAZE_p_TCGASNP_b86_87_88_N_GenomeWideSNP_6_C08_735516  1   61735   12785720    6509    0.0628
AMAZE_p_TCGASNP_b86_87_88_N_GenomeWideSNP_6_C08_735516  1   12792599    13278863    57  -0.5061
AMAZE_p_TCGASNP_b86_87_88_N_GenomeWideSNP_6_C08_735516  1   13326062    25256850    7563    0.0667
AMAZE_p_TCGASNP_b86_87_88_N_GenomeWideSNP_6_C08_735516  1   25266637    25336819    32  0.6428
AMAZE_p_TCGASNP_b86_87_88_N_GenomeWideSNP_6_C08_735516  1   25336853    57755402    18550   0.0636
AMAZE_p_TCGASNP_b86_87_88_N_GenomeWideSNP_6_C08_735516  1   57757884    69368291    8418    0.1007
AMAZE_p_TCGASNP_b86_87_88_N_GenomeWideSNP_6_C08_735516  1   69375800    84595696    9774    0.0659

[poisonAlien]

I believe that the papers that you're mentioning are looking at population based cnv's rather than sample wise. TCGA uses gistic to identify recurrent CN aberration in a cohort, which are potentially driver events.

Above file is a CBS output and not all entries are copy number alterations. Last column Segment_Mean shows how much of a genomic segment differs from normal in log2 scale.