使用rpy2设置晶格图选项的问题

Question

I'm trying to create a heatmap or color-intensity plot using data from a numpy array, using rpy2 and lattice. I'm using python 2.6.2, R 2.10.1, rpy2 2.1.9, not sure which version of lattice. I've gotten it working perfectly, except that I need to modify the default lattice setting for the color ramp used to plot the levels of the relevant variable (z). Specifically, I want grayscale instead of the magenta-cyan default ramp. Here is code to generate a dummy dataframe and create the grayscale levelplot in vanilla R:

library(lattice)

x <- rep(seq(1,10), each=10)
y <- rep(seq(1,10), 10)
z <- abs(rnorm(100))
z <- z/max(z)
df <- data.frame(x=x, y=y, z=z)

grayvector <- gray(seq(0,1,1/100))

foo <- levelplot(z ~ x * y, data=df, col.regions = grayvector)
print foo

With rpy2, I cannot set the col.regions argument. According to the documentation, rpy2 is supposed to convert any . characters in function arguments to _ . This doesn't appear to be working, however, since using col_regions results in the argument being ignored. Here is the python code that produces the levelplot, but without grayscale:

from __future__ import division
import rpy2.robjects as ro
from rpy2.robjects.packages import importr
r = ro.r
lattice = importr("lattice")

grayvector = r.gray( r.seq(0, 1, 1/100))   
x = r.rep(r.seq(1,10), each=10)
y = r.rep(r.seq(1,10), 10)
z = r.abs(r.rnorm(100))

df = {'x': x, 'y' :y, 'z':z}
df = ro.DataFrame(foo)

formula = ro.Formula('z ~ x * y')
formula.getenvironment()['z'] = df.rx2('z')
formula.getenvironment()['y'] = df.rx2('y')
formula.getenvironment()['z'] = df.rx2('z')

foo = lattice.levelplot(formula, data=df, col_regions = grayvector)
print foo

Does anyone know how to use lattice function arguments with a . in them in rpy2?

Answer 1

You need to specify the argument mapping manually:

from rpy2.robjects.functions import SignatureTranslatedFunction
lattice = importr("lattice")
lattice.levelplot = SignatureTranslatedFunction(lattice.levelplot,
                                                init_prm_translate={'col_regions': 'col.regions'})
foo = lattice.levelplot(formula, data=df, col_regions=grayvector)

And also check this: http://rpy.sourceforge.net/rpy2/doc-2.2/html/robjects_functions.html

It is important to understand that the translation is done by inspecting the signature of the R function, and that not much can be guessed from the R ellipsis '...' whenever present.