Page 8 - Cancer Systems Biology: Methods and Protocols

_.....~.....a
a
4 Hua Tan and Xiaobo Zhou "
Tumor Samples
E:/IHI \IHTL"

z LL ATTTH:aw

[om [wad roe

Fig. 1 Schematic representation of two combinatorial mutational patterns studied in this protocol: the
co-mutational pattern (upper pane) refers to the scenario that a set of genes tends to mutate simultaneously
in a tumor sample, whereas the mutually exclusive pattern (lower panel represents the opposite scenario:
genes in a given set tend to avoid mutating simultaneously in any one tumor sample

[ 5]. Previous experimental and statistical analyses have consistently
revealed two combinatorial mutational patterns for a given set of
genes, termed co-mutational and mutually exclusive patterns
[5, 8-10]. As shown in Fig. 1, the co-mutational pattern occurs
when a set of genes tend to mutate simultaneously in a single
tumor, while the mutually exclusive pattern refers to the scenario
in which one and only one of a set of genes is likely to be altered in a
tumor.
Mutually exclusive genes are likely to function in the same
signaling pathway, whereas co-mutational genes are likely to take
effect in different pathways [11]. Combinatorial patterns of genes
can be leveraged to infer signaling networks implicated in human
cancer development and progression. Indeed, many efforts have
been devoted to de novo discover novel driver pathways based on
mutual exclusivity of gene mutations [ 11-13]. Therefore, it has
essential biological relevance to identify gene pairs or gene sets with
significant combinatorial mutational patterns.
Previous work proposed a statistical method to deal with this
question and nominated a number of gene sets with significant
combinatorial patterns [10]. However, this analysis was performed
on a batch of very limited cell line data. The analysis thus lacks an
elaborate procedure to preprocess data from a giant mutation
database which consists of a large number of clinical samples of
various cancer types ( e.g., the recently released Catalog of Somatic
Mutations in Cancer COSMIC [14] and the Cancer Genome
Atlas TCGA, https://tcga-data.nci.nih.gov /tcga/). In addition,
the analysis by Yeang et al. adopted different hypothesis tests to
estimate the significance levels of the two combinatorial mutational
patterns, which tend to yield a too conservative p-value for the
co-mutational pattern [10].
To address these issues, we here describe a systematic and reliable
pipeline to identify both combinatorial mutational patterns in cancer
genomes. Here, somatic mutations exclude the synonymous point