Fig 1.
The causal structure of the mutation in the cancer biological process.
M is a binary treatment variable, i.e., whether a gene mutated or not; Y is the outcome, e.g., the cell proliferation activity. X is the observed confounders, i.e., the mutation matrix of covariant genes; Z is the unobserved confounders which can be learned from X, e.g., micro-environment stress, so another role for X is a proxy variable for Z; Confounder affect both the treatment and the outcome and may lead to the erroneous association.
Fig 2.
An illustration of the proposed framework.
The biological process activity of each sample is obtained by core genes regression and used as the outcome of the casual model. The casual model consists of an inference network to calculate the mean and variance of P(Z|X,y,m) and a model network to recover P(Z,X,y,m).
Fig 3.
Top 10 mutations with the highest causal effect on DNA replication in 10 cancers.
The figure shows the ATE value and the mutation rate of each predicted key mutation.
Fig 4.
Top 10 mutations with the highest causal effect on EMT process in 10 cancers.
The figure shows the ATE value and the mutation rate of each predicted key mutation.
Table 1.
The causal effect of the top 10 genes with highest mutation rates on DNA replication.
Table 2.
The causal effect of the top 10 genes with highest mutation rates on EMT.
Fig 5.
The Mann-Whitney U test (M.W.) analysis reflects the significant differences between mutated groups and non-mutated groups of mutated genes with high ATE values in the activity of DNA replication (DR) for 10 cancers.
Fig 6.
The Mann-Whitney U test (M.W.) analysis reflects the significant differences between mutated groups and non-mutated groups of mutated genes with high ATE values in the activity of EMT process for 10 cancers.
Fig 7.
The Mann-Whitney U test (M.W.) analysis shows there is no significant differences between mutated groups and non-mutated groups of mutated genes with high mutation rates in the activity of DNA replication (DR) for 10 cancers.
Fig 8.
The Mann-Whitney U test (M.W.) analysis shows there is no significant differences between mutated groups and non-mutated groups of mutated genes with high mutation rates in the activity of EMT process for 10 cancers.
Table 3.
The statistics of the cancer datasets used in this study.