Using IsoplotR in iolite

Joe

Discussion for note.

tonydoss

I've tried installing rpy2 with the instructions provided here https://notes.iolite.xyz/posts/installingpackages/... but rpy2 doesn't seem to be provided with a .whl file: https://pypi.org/project/rpy2/#files (and the instructions seem to only apply to a .whl file renamed to .zip)
So I've installed rpy2 from my python command prompt, and from Spyder (to be sure)... but Iolite doesn't seem to see the package. Even after restart iolite, when I try to run the line "from rpy2 import robjects as ro" in the python workspace, I get "No module named 'rpy2'"
(note that compiling the measurements worked worked fine)

Joe

tonydoss

You can use the .tar.gz from pypi for rpy2. It is a compressed file just like a zip. You should be able to extract it on a Mac with the built in tools. On PC 7zip should do the trick. Then you would copy the rpy2 and rpy2.egg-info folders from the archive to the site-packages folder that iolite uses.

iolite uses its own python distribution and site-packages so installing a package via pip or Spyder will not put it in the correct place. If the version of python on your computer matches that in iolite (v3.6 at the moment), you could install via command line/Spyder, find where those files were installed, and copy them to iolite's site-packages path.

All the best,

tonydoss

Thanks. I move the rpy2 and rpy2.egg-info folders to C:\Program Files (x86)\iolite-software\iolite4\python3.6\site-packages...
but now I get:
Error Type: <class 'ModuleNotFoundError'>
Error Value: No module named '_rinterface_cffi_abi'
Error Traceback: File ""<string>"", line 1, in <module>

mmm. indeed I can't see any _rinterface_cffi_abi.py in the rpy2 folder.

Joe

tonydoss

You're right, the rpy2 installation is bit more involved. I thought it was pure python, but the build scripts generate some other files. The required site-packages can be downloaded here for windows. Unzip the contents to the site-packages and restart iolite. I've tested this on clean install and it seemed to work. R needs to be installed separately. I tested with R version 4.0.2. Rpy2 requires an 'RHOME' environment variable is set so it can locate your R installation. This should be setup automatically by the R installer. In the event that you have more than 1 R installation and/or it isn't set, you could do something like:

import os
os.environ['RHOME'] = '...'

where '...' is replaced with the path to R and these commands are run before you import from rpy2.

Note that rpy2 has changed a bit since this note was written. The middle section should look something like below now, where the python/pandas data frame has been reordered the way IsoplotR expects and the conversion is via py2rpy rather than py2ri.

df = pd.DataFrame({
	'x': r75,
	'sx': r75s,
	'y': r68,
	'sy': r68s,
	'rho': rho
})
df = df[ ['x', 'sx', 'y', 'sy', 'rho'] ]

with localconverter(ro.default_converter + pandas2ri.converter):
	rdf = ro.conversion.py2rpy(df)

tonydoss

Joe
Thanks. It seems it can find R without needing os.environ (I got R 3.6.0, I'm a bit reluctant to update to 4 because of package compatibility issues), but I get this:
Error Type: <class 'OsError'>
Error Value: cannot load library 'C:\Program Files\R\R-3.6.0\bin\x64\R.dll': error 0x7e
Error Traceback: File ""<string>"", line 1, in <module>

note I'm more interested in Th230-U dating than U-Pb, but I guess if I can make it work for U-Pb I should be able to make it work for Th230-U 😀

Joe

tonydoss

Yes, I'm afraid the packages I uploaded are tied to R 4.0.

tonydoss

Joe no worries, I've updated to R 4.0 (nervously... and rightfully, because now I have to re-install a whole bunch of packages... no idea how it didn't transfer them)
I'm still getting the same error: cannot load library 'C:\Program Files\R\R-4.0.2\bin\x64\R.dll': error 0x7e

Joe

tonydoss

Can you report the output of running the following in iolite's python workspace:

from rpy2 import situation
import sys
sys.argv = ['iolite']
for ii in situation.iter_info():
	print(ii)

from rpy2 import robjects as ro
a = ro.reval('a=1')
print(a[0])

For me the on a PC with only R4.0.2 installed and nothing else related to R, I see:

rpy2 version:
3.3.5
Python version:
3.6.3 (v3.6.3:2c5fed8, Oct  3 2017, 18:11:49) [MSC v.1900 64 bit (AMD64)]
Looking for R's HOME:
    Environment variable R_HOME: None
    InstallPath in the registry: C:\Program Files\R\R-4.0.2
    Environment variable R_USER: None
    Environment variable R_LIBS_USER: None
R version:
    In the PATH: 
    Loading R library from rpy2: OK
Additional directories to load R packages from:
None
C extension compilation:
    Warning: Unable to get R compilation flags.
1.0

tonydoss

Joe
This is what I get...

rpy2 version:
3.3.5
Python version:
3.6.3 (v3.6.3:2c5fed8, Oct  3 2017, 18:11:49) [MSC v.1900 64 bit (AMD64)]
Looking for R's HOME:
    Environment variable R_HOME: None
    InstallPath in the registry: C:\Program Files\R\R-4.0.2
    Environment variable R_USER: None
    Environment variable R_LIBS_USER: None
R version:
    In the PATH: None
    Loading R library from rpy2: cannot load library 'C:\Program Files\R\R-4.0.2\bin\x64\R.dll': error 0x7e
Additional directories to load R packages from:
None
C extension compilation:
    Warning: Unable to get R compilation flags.

    Traceback (most recent call last):
  File "", line 7, in 
  File "", line 971, in _find_and_load
  File "", line 955, in _find_and_load_unlocked
  File "", line 656, in _load_unlocked
  File "", line 626, in _load_backward_compatible
  File "C:/Program Files (x86)/iolite-software/iolite4/python3.6/site-packages\rpy2/robjects\__init__.py", line 16, in 
   
import rpy2.rinterface as rinterface
  File "", line 971, in _find_and_load
  File "", line 955, in _find_and_load_unlocked
  File "", line 656, in _load_unlocked
  File "", line 626, in _load_backward_compatible
  File "C:/Program Files (x86)/iolite-software/iolite4/python3.6/site-packages\rpy2/rinterface.py", line 6, in 
 
from rpy2.rinterface_lib import openrlib
  File "", line 971, in _find_and_load
  File "", line 955, in _find_and_load_unlocked
  File "", line 656, in _load_unlocked
  File "", line 626, in _load_backward_compatible
  File "C:/Program Files (x86)/iolite-software/iolite4/python3.6/site-packages\rpy2\rinterface_lib/openrlib.py", line 44, in 
 
rlib = _dlopen_rlib(R_HOME)
  File "C:/Program Files (x86)/iolite-software/iolite4/python3.6/site-packages\rpy2\rinterface_lib/openrlib.py", line 37, in _dlopen_rlib
 
rlib = ffi.dlopen(lib_path)

OSError
: 
cannot load library 'C:\Program Files\R\R-4.0.2\bin\x64\R.dll': error 0x7e

bob

You ever figure out the "OSError: cannot load library 'C:\Program Files\R\R-3.6.0\bin\x64\R.dll': error 0x7e" error? Im having the same problem and am quite stumped.

tonydoss

bob nope (I gave up)... @Joe , any ideas?

Joe

I'm afraid I do not have a solution @bob and @tonydoss. I just tried it again on a fresh Windows 10 virtual machine with the latest R downloaded and installed with all the default settings and it works fine.

It is perhaps worth noting that the rpy2 repository has a relatively recent unresolved issue on this particular topic that can be found here:
https://github.com/rpy2/rpy2/issues/721

There is something about your particular setups that the python + R + rpy2 combination does not like, but it is not clear what it is.

I tried installing an older version of R in my virtual machine and it continued to work initially. Then after rebooting (maybe?) I did get the quoted error! I then uninstalled 3.6.3 and got a different error related to r_home not being set. So I then set r_home as below:

import os
os.environ['r_home'] = 'C:\\Program Files\\R\\R-4.0.3'

and it worked again. Then I tried to recreate the error by installing 3.6.3 again, but it continued to work. So, I think it is probably related to rpy2's ability to detect your R installation when multiple versions are installed. You could try setting the environment variable r_home as above (replacing the version code/path as appropriate) before trying to use R in python, but I'm not sure that it'll work.

All the best,

tonydoss

Joe I've tried the commands you suggested (trialing two different installs of R I have on my laptop - 3.6.0 and 4.0.2) but still get the same error message in either case (can't load R.dll library). Bummer.

hourigan

This is great and really helpful. I am attaching a doctored screen shot of the packages that I needed to copy from the site-packages directory of my installation of Python 3.8 into Iolite's Python 3.8 site packages

For rpy2 to find my installation of R 4.2 on a OSX Catalina, I used:

import os
os.environ['R_HOME'] = '/Library/Frameworks/R.framework/Versions/4.2/Resources'

I am definitely able to generate concordia ages and concordia plots with one modest edit:

with localconverter(ro.default_converter + pandas2ri.converter):
    rdf = ro.conversion.py2ri(df)

needed to be:

with localconverter(ro.default_converter + pandas2ri.converter):
    rdf = ro.conversion.py2rpy(df)

I am facing the problem that is highlighted for the Linux system that the Quartz window that is generated cannot be closed without crashing Iolite. Has anyone played around with generating a PDF without opening a quartz window?
Clearly absurd to generate a concordia age on detrital sample

Joe

hourigan

I can confirm that I also needed to make these changes for the latest versions of R, IsoplotR, and rpy2 on Windows. On windows, I set the R_HOME path as:

os.environ['R_HOME'] = 'C:\Program Files\R\R-4.2.0'

I was able to produce a pdf following the rpy2 guidance on non-interactive plotting found here, which boils down to:

grdevices = importr('grDevices')
grdevices.pdf(file="/Path/To/File.pdf")
ir.concordia(ird, show_age=1)
grdevices.dev_off()

I.e. wrapping the plotting commands with the grdevices.pdf(...) and grdevices.dev_off() calls. Other filetypes are possible by replacing .pdf with, e.g., .png.

All the best,

hourigan

Joe

I started to try to work up the isoplotR code with PDF generation rather than opening a quartz (I guess this is iOS specific?) window.

I am getting a 'failed to load default encoding' error. Before I go down the rabbit hole, I was wondering if this is something that you had seen in working up the example above

Here is the Python Consule output of the error:
_R[write to console]: Error in (function (file = if (onefile) "Rplots.pdf" else "Rplot%03d.pdf", :
failed to load default encoding

R[write to console]: In addition:
R[write to console]: Warning message:

R[write to console]: In (function (file = if (onefile) "Rplots.pdf" else "Rplot%03d.pdf", :
R[write to console]:

R[write to console]: failed to load encoding file 'ISOLatin1.enc'

Traceback (most recent call last):
File "/Volumes/GoogleDrive/My Drive/Iolite/Plugins/User interface/ioPlotR_UPb.py", line 222, in isoplotConcordia
grdevices.pdf(file = pdf_path)
File "/Applications/iolite4.app/Contents/MacOS/../Resources/python3.8/rpy2/robjects/functions.py", line 201, in call
return (super(SignatureTranslatedFunction, self)
File "/Applications/iolite4.app/Contents/MacOS/../Resources/python3.8/rpy2/robjects/functions.py", line 124, in call
res = super(Function, self).call(new_args, **new_kwargs)
File "/Applications/iolite4.app/Contents/MacOS/../Resources/python3.8/rpy2/rinterface_lib/conversion.py", line 45, in _
cdata = function(args, **kwargs)
File "/Applications/iolite4.app/Contents/MacOS/../Resources/python3.8/rpy2/rinterface.py", line 810, in call
raise embedded.RRuntimeError(rinterface.geterrmessage())
rpy2.rinterface_lib.embedded.RRuntimeError: Error in (function (file = if (onefile) "Rplots.pdf" else "Rplot%03d.pdf", :
failed to load default encoding_

hourigan

Thanks for that guidance Joe!

I put together an export script that formats results into isoplotR's json file for direct import of U-Pb data into IsoplotR.

Produces a *.json file for each selection group (i.e. sample)
Reads in a "json" template that was generated by saving data in isoplotR to a JSON file.
Lines 49 -143 allow customization of IsoplotR's U-Pb settings for particular types of samples (improved reproducibility of plot style, axes etc?)
Writes 7/35, 6/38, and 7/6 ratio data to the JSON file
5 Line 12: User can specify whether to export internal or propagated uncertainties
6 IsoplotR Error ellipse color: comment / uncomment lines 44,45 or 46 to specify 204Pb_CPS, 238U_CPS or Raw_U/Th or make your own.
Edit Line 22 to the directory you want to save your JSONs in.
Edit Line 18 to be your JSON template path. I've found that the python JSON reader has problems with the full IsoplotR JSON file. I edited my template to include only the 'constants', 'settings', and 'data' items. I removed 'languages', 'data4server', and 'optionschanged'

Attachments 1 : Export script
(I wanted to inline the code in this post, but # comments end up as heading sized text.)

isoplotr-upbjson-forumn.txt

6kB

Attachment 2: JSON Template

template.json

23kB

Joe

hourigan

Thanks for sharing!

PS: You can post code by wrapping it in single (in-line) or triple (block) backticks (`) or by clicking the </> button in the toolbar. Since the editor recognizes markdown, code not wrapped in the "code" markers is parsed as such, which is why the text preceded by '#' is rendered as a heading.

hourigan

@Bence @Joe

I don't know what was going on with the encoding. I am not sure that I did anything to fix it.
Making good progress. I can definitely offload "Final_Ratios" for calculation of S-K corrected PbCorrect 206/238 ages and I am now successfully plotting IsoplotR concordia to PDF format.

In the attached video, none of the plotting options are live. In other words, just the default settings are being used (Wetherill concordia, green / red ellipse coloring and no color channel) but these are details. The big hurdle was definitely the PDF plotting, which remains admittedly buggy.

I am going to next the IoPlotR functionality on the menu to keep the widgets simpler.
Tools
-->IoPlotR
---->Concordia
---->Ages
---->Weighted Mean
---->KDE

https://youtu.be/PP8-oU4zPAQ