A task-oriented R interface to the 'RDKit' < https://www.rdkit.org> library through its 'Python' API via 'reticulate'. The package offers high-level cheminformatics functionality, including molecule parsing, descriptor calculation, and fingerprint generation without replicating the native structure of 'RDKit'.
The rdkitpyr package provides a high-level interface for cheminformatics
tasks in R. It leverages the RDKit library via its Python API, accessed through
the reticulate package. Unlike a thin wrapper around RDKit, rdkitpyr offers
R-friendly, task-oriented functions without mirroring RDKit's native API. The
package motto is "Focus on chemistry tasks in R, without diving into RDKit
internals".
The name rdkitpyr (pronounced /ar-dee-kit-peer/) highlights its nested
structure (from R to Python to the RDKit C++ library) and its role as an R-side
peer to RDKit rather than a direct wrapper.
A key design principle is one Python call per task. Custom Python helper functions implement this approach, reducing overhead and ensuring consistent exception handling.
# Install 'rdkitpyr' from CRAN:
install.packages("rdkitpyr")
# Install 'rdkitpyr' from GitHub:
library(devtools)
install_github("andreysamokhin/rdkitpyr")
The rdkitpyr package supports both novice and advanced users.
Python>=3.12,
numpy>=2,<3, and rdkit>=2025.9.1) and select it in R using one of the
methods recommended by reticulate (e.g., setting the RETICULATE_PYTHON_ENV
environment variable or using reticulate::use_*() functions). This gives
full control over package versions and the Python environment.reticulate using a manifest created with
reticulate::py_require(). Installation starts only when a function that
requires Python is called. It may take a few minutes, and requires about 200 MB
of disk space.Here is a simple example:
library(rdkitpyr)
rdkitpyr::ConvertToInchi(c("CC", "CCC"))
#> "InChI=1S/C2H6/c1-2/h1-2H3"
#> "InChI=1S/C3H8/c1-3-2/h3H2,1-2H3"
rdkitpyr::CalculateExactMass(c("c1ccccc1", "CCCC"))
#> 78.04695 58.07825