rdiaz02 / evam-tools Goto Github PK

• Create data interactively
• Add genotypes, WT, E.
• Then decrease gene number: crash

Those are needed to get the likelihood and compute cross-validated likelihoods, etc. Right now, the predicted genotypes are those for models without noise.

• H-ESBCN returns the best epsilon (see function markov.h)
• CBN: see line 286 in h-cbn.c : write the epsilon and lik. to a file.

This works for DAGs

plot_model <- function(x, main = "") {
    method_info <- igraph::graph_from_data_frame(x[, c("From", "To")])
    parent_set <- evamtools:::parent_set_from_edges(x)
    evamtools:::plot_method(method_info = method_info,
                            parent_set = parent_set,
                            edges = x, 
                            method = main)
  }

where is the usual data frame with From, To, etc.

Improve and do something similar for MHN.

The above is simplified from plot_evam in file evam_plotting.R

A user might try this

d1 <- data.frame(From = c("Root", "Root", rep(c("A", "B"), 3)),
                              To = c("A", "B", rep(c("C", "D", "E"), c(2, 2, 2))),
                              Relation = rep(c("Single", "AND", "OR", "XOR"), rep(2, 4)),
                              Lambdas = c(0.7, 0.8, rep(c(0.9, 0.4, 0.5), c(2, 2, 2))))

evamtools:::cpm2tm(d1)

which fails as

Error in as.environment(where) : using 'as.environment(NULL)' is defunct
> traceback()
2: exists("Relation", x$edges)
1: evamtools:::cpm2tm(ex_XOR_AND_OR)

So we probably want to:

1. Check for the existance of the edges component and error early and with a clear message (like "No edges component in input. You are calling this function in an unintended way and/or with the wrong input").
2. Use "Relation" %in% colnames(x$edges) instead of exists("Relation", x$edges): the second fails when x$edges does not exist.

• Create input data interactively
• Add WT, E, D genotypes
• Rename data
• Move to "Empty" (from which it was derived): CRASH

Many variations on this. In general, moving between data sets when some have genes that are not consecutive.

This currently does not work

d1 <- data.frame(From = c("Root", "Root", rep(c("A", "B"), 3)),
                              To = c("A", "B", rep(c("C", "D", "E"), c(2, 2, 2))),
                              Relation = rep(c("Single", "AND", "OR", "XOR"), rep(2, 4)),
                              Lambdas = c(0.7, 0.8, rep(c(0.9, 0.4, 0.5), c(2, 2, 2))))

sample_evam(d1)

And yet, the web app is doing something that seems to be just that.

So:

1. We should allow for that type of use in the R package
2. We should document it in the help of sample_evam
3. HESBCN_model_2_output should accept an empty parent_set argument, and generate it on the fly from the dag (we already have function parent_set_from_edges).

In the web app we are using function generate_sample_from_dag. (Called on click on "resample_dag")

As it says, bubble plots as in Moen & Johnston, HyperHMM: efficient inference of evolutionary and progressive dynamics on hypercubic transition graphs, https://academic.oup.com/bioinformatics/article/39/1/btac803/6895098 , Fig.2, 3, o 4, left column