Mining Affymetrix SNP6.0 copy number data in human cancer
Adam Butler
Morning Session, 2 September (11th MGED Meeting, 1-4 September, 2008)
They are looking for homozygous deletions, genomic amplifications, regions of LOH, and want to map break points. They have 1246 samples to do this work on.
The platform
906,600 SNPs, with 946,000 copy number probes. But how well does it work in the real world? They plotted out the results, and saw nice, w ell-defined datapoints with good resolution for the SNPs. The copy number probes also looked good. Combining the two data sets also went well.
Segmentation
Segmentation was another way to view and analyze the data (segmentation analysis). Original analysis had a problem with triploid genes (which often happened with cancer) and other amounts of copy numbers in complex genomes. Therefore instead they started using ratios rather than absolute copy numbers.
Mining
CONAN is their copy number analysis web tool.
Preliminary Results
There are ~7500 homozygous deletions, 6182 in the cancer cell lines and 1341 in the "normals". There were 2891 putatively somatic homozygous deletions which group into 968 clusters. Of these clusters, 902 are of unknown significance, of which 716 are singletons.
http://www.sanger.ac.uk/genetics/CGP
These are just my notes and are not guaranteed to be correct.
Please feel free to let me know about any errors, which are all my
fault and not the fault of the speaker. :)
