Hi Everyone, We have a question about whether one can to do a particular type of Granger Causality (GC) network validation in R. We hope you'll agree it's an interesting problem and that someone's figured out how to solve it. We have a cellular network with n nodes (proteins).? We have two different n x s x k time series matrices that describe the network activity under two mutually exclusive conditions, C (cancerous cell) and H (healthy cell), where s is the length of the time series data, and k is the number of observations. Using the time series matrices, we calculated two different n x n GC matrices, one for healthy cells and one for cancerous cells, so that ij th element in each matrix represents the GC influence of node i on node j.? Using the various standard tests, we know that many of the GC values are extremely significant. Now we?re given a brand-new observation in the form of a n x s x 1 time series matrix Y that represents the activity of the same n nodes (we don?t know a priori whether the new data come from a healthy cell or a cancerous cell). Given this matrix Y : (1) How can we go about determining if Y comes from a cancerous cell (condition C) or a healthy cell (condition H)? (2) Is there a package in R that we can use for this purpose? Thank you very much! Pat
Network validation (of sorts) using granger Causality in R
3 messages · PWD7052, Erdogan CEVHER, Bert Gunter
1. Describe better the distributions of obs to time series; i.e., describe clearly the time series and obs with math'l notation briefly. 2. Use Conditional G-causality and/or partial G-causality, you can exceed "the limit of max. number of variables = 11" in a VAR structure. 3. You can use R's FIAR package for CGC / PGC calculations. I advise version 3 of that package. 4. Kamamoto Oscillators for "Graphical" GC methods perhaps may well suit to your case. 5. Matlab's GCCA and MVGC packages cannot handle cointegration (in case cointegrated vars exist) whereas you can handle cointegrated cases via CGC and PGC. 2017-01-03 9:20 GMT+03:00 PWD7052 via R-help <r-help at r-project.org>:
Hi Everyone,
We have a question about whether one can to do a particular type of
Granger Causality (GC) network validation in R. We hope you'll agree it's
an interesting problem and that someone's figured out how to solve it.
We have a cellular network with n nodes (proteins). We have two different
n x s x k time series matrices that describe the network activity under two
mutually exclusive conditions, C (cancerous cell) and H (healthy cell),
where s is the length of the time series data, and k is the number of
observations.
Using the time series matrices, we calculated two different n x n GC
matrices, one for healthy cells and one for cancerous cells, so that ij th
element in each matrix represents the GC influence of node i on node j.
Using the various standard tests, we know that many of the GC values are
extremely significant.
Now we?re given a brand-new observation in the form of a n x s x 1 time
series matrix Y that represents the activity of the same n nodes (we don?t
know a priori whether the new data come from a healthy cell or a cancerous
cell).
Given this matrix Y :
(1) How can we go about determining if Y comes from a cancerous cell
(condition C) or a healthy cell (condition H)?
(2) Is there a package in R that we can use for this purpose?
Thank you very much!
Pat
[[alternative HTML version deleted]]
______________________________________________ R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/ posting-guide.html and provide commented, minimal, self-contained, reproducible code.
Have you searched?! "Granger causality" at rseek.org brought up what appeared to be many relevant hits. -- Bert Bert Gunter "The trouble with having an open mind is that people keep coming along and sticking things into it." -- Opus (aka Berkeley Breathed in his "Bloom County" comic strip ) On Mon, Jan 2, 2017 at 10:20 PM, PWD7052 via R-help
<r-help at r-project.org> wrote:
Hi Everyone,
We have a question about whether one can to do a particular type of Granger Causality (GC) network validation in R. We hope you'll agree it's an interesting problem and that someone's figured out how to solve it.
We have a cellular network with n nodes (proteins). We have two different n x s x k time series matrices that describe the network activity under two mutually exclusive conditions, C (cancerous cell) and H (healthy cell), where s is the length of the time series data, and k is the number of observations.
Using the time series matrices, we calculated two different n x n GC matrices, one for healthy cells and one for cancerous cells, so that ij th element in each matrix represents the GC influence of node i on node j. Using the various standard tests, we know that many of the GC values are extremely significant.
Now we?re given a brand-new observation in the form of a n x s x 1 time series matrix Y that represents the activity of the same n nodes (we don?t know a priori whether the new data come from a healthy cell or a cancerous cell).
Given this matrix Y :
(1) How can we go about determining if Y comes from a cancerous cell (condition C) or a healthy cell (condition H)?
(2) Is there a package in R that we can use for this purpose?
Thank you very much!
Pat
[[alternative HTML version deleted]]
______________________________________________ R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.