Theoretical Systems Biology Retreat
(German Cancer Resarch Center, Heidelberg)
Ellwangen · 22 June 2016
▷ Learn about gene regulation?
Model fitting
Granger Causality/Transfer Entropy
Causal Inference Methods
Convergent Cross Mapping
Model fitting
Granger Causality/Transfer Entropy
Causal Inference Methods
Convergent Cross Mapping
Conditional independence tests: Xti⊥⊥X(t−1)j|X(t−1)k1,X(t−1)k2,...
Model fitting
Granger Causality/Transfer Entropy
Causal Inference Methods
Convergent Cross Mapping
Idea: Reconstruct regulating gene from the history of the regulated gene.
Idea: Reconstruct regulating gene from the history of the regulated gene.
▷ Can exploit branching geometry!
▷ But: no statistical formulation!
Rationale
Xtj=fj(X(t−1)j,X(t−1)i1,X(t−1)i2,...)⇔Fj(Xtj,X(t−1)j,X(t−1)i1,X(t−1)i2,...)=0Statistically test whether X(t−1)i1 contributes to Fj by testing for the existence of a function g g(Xtj,X(t−1)j,X(t−1)i2,...)=X(t−1)i1.
Implicit function theorem: if there is no g, Fj is constant w.r.t X(t−1)i1.
Consider Xtj=fj(X(t−1)j,X(t−1)i). How to statistically test for the existence of mapping the g:Xj→Xi?
zji=¯Δji−¯δi√(ˆσΔji)2/n+(ˆσδji)2/n where Δ(a)tji=1|N(a)tj|∑t′∈N(a)tjdtt′i is the variation across realizations for a fixed value in the domain Xj δ(a)ti=12(d(a)t(t−1)i+d(a)(t+1)ti) is the dynamic variation in the codomain.