Analysis of array CGH data for the estimation of genetic trumor progression

Progression
Abstract
In cancer research, prediction of time of death or relapse is important for a meaningful
tumor classification and selecting appropriate therapies. The accumulation of genetic
alterations during tumor progression can be used for the assessment of the genetic
status of the tumor. For modeling dependences between the genetic events,
evolutionary tree models have been used ([1], [2]). ArrayCGH data is now available
for various types of gliomas ([3]). Our goal is to estimate GPS trees for glioblastomas,
by identifying specific chromosomal alterations in this arrayCGH data.
Many good algorithms have been proposed for estimating the copy number of DNA
regions from noisy arrayCGH data. We chose GLAD ([4]), which uses an adaptive
weights smoothing algorithm to detect breakpoints of segments with constant copy
number. We also tried Quantreg ([5]) and DNAcopy (?), with similar predictions.
Once we obtain the piecewise constant functions that estimate the copy numbers, a
threshold is chosen to decide whether a region is normal, lost or gained. This
threshold is specific for each sample, and it is computed as the 95 percentile of a
robust normal fitted to the frequency distribution of the estimated copy numbers.
Computing the genetic progression trees require the detection of the most significant
regions across all samples. We propose an algorithm that first selects highly
statistically significant regions and then extends these regions based on a similarity
measure. The results consist of both isolated genes involved in oncogenesis and wide
regions (the whole chromosome 10, or whole arm of chromosome 9). The estimated
genetic trees are consistent with previous observations on glioblastomas genetic
progression.
References:
[1] Estimating cancer survival and clinical outcome based on genetic tumor
progression scores – Joerg Rahnenfuehrer, Niko Beerenwinkel, et. al., 2005
[2] Learning Multiple Evolutionary Pathways from Cross-Sectional Data - Joerg
Rahnenfuehrer, Niko Beerenwinkel, et. al., 2005
[3] High-Resolution Genome-Wide Mapping of Genetic Alterations in Human Glial
Brain Tumors – Markus Bredel, et.al., 2005
[4] Analysis of array CGH data: from signal ratio to gain and loss of DNA regions –
Philippe Hupe et. al. , 2004
[5] Quantile smoothing of array CGH data - Paul H.C. Eilers et. al., 2004