Installation

msbuddy requires Python version 3.9+. There are a variety of ways to install msbuddy.

Option 1: Using pip (recommended)

msbuddy is available on PyPI, you can install it via pip:

pip install msbuddy

To perform formula annotation, you also have to install the lightgbm package.

pip install lightgbm

Option 2: From source

You can also install msbuddy from source:

git clone https://github.com/Philipbear/msbuddy.git

Option 3: From PyPI release

Alternatively, you can download the latest release from PyPI and install it:

tar -zxvf msbuddy-0.3.0.tar.gz

Data files

For backwards compatibility, msbuddy still uses the msbuddy/data directory beside the installed package when no data options are supplied.

For a system installation, a writable directory can instead be selected in the existing configuration object:

from msbuddy import Msbuddy, MsbuddyConfig

config = MsbuddyConfig(data_dir="~/msbuddy-data")
engine = Msbuddy(config)

Compatible files can also use different names. Relative names are resolved inside data_dir and absolute paths are used directly:

config = MsbuddyConfig(
    data_dir="~/msbuddy-data",
    data_files={
        "formula_db": "formula-custom.joblib",
        "ml_model": "/shared/msbuddy/model-custom.joblib",
    },
)

The supported keys are common_db, formula_db, and ml_model. A partial mapping is allowed; unspecified files keep their standard names.

A custom file that already exists is loaded as-is. A missing custom file raises a clear error instead of silently substituting another file. To place the standard release asset under a custom name, call download_data explicitly.

The file paths and standard download links can be inspected, or all files can be downloaded in advance:

from msbuddy import download_data, get_data_files

print(get_data_files(config.data_dir, config.data_files))
download_data(config.data_dir, config.data_files)

The existing command-line interface provides matching options. Examples use the conventional double-dash form for descriptive option names.

Existing single-dash spellings such as -mgf remain available as aliases for backwards compatibility.

msbuddy --download-data --data-dir ~/msbuddy-data
msbuddy --mgf input.mgf --data-dir ~/msbuddy-data \
    --formula-db formula-custom.joblib \
    --ml-model /shared/msbuddy/model-custom.joblib

Omitting all new options preserves the original location and automatic download behavior. Permission errors name the failed path and show the writable-directory alternatives.