Python API — Overview¶

The fastdpplot Python package is a thin layer on top of the Rust extension (fastdpplot._rs). All heavy lifting — file I/O, binning, sparse conversion — happens in Rust. Python adds:

Friendly function signatures with sensible defaults.
A pandas DataFrame interface consumed by the rendering stack.
A datashader + Panel interactive server.
Optional cuDF GPU acceleration.

Module summary¶

Module	Public symbols	Purpose
`io`	`load()`, `load_bin()`	Load data from Parquet, text, or `.bin`+FASTA.
`plot`	`render_static()`, `serve_interactive()`, `show()`	Render or serve dotplots.
`server`	`run()`, `run_bin()`	Start the Panel server from Python.
`sparse_convert`	`to_scipy_sparse()`	Convert binning output to a scipy sparse matrix.

Import conventions¶

# High-level: recommended for most users
from fastdpplot.io import load, load_bin
from fastdpplot.plot import render_static, serve_interactive, show

# Low-level: access the Rust extension directly
from fastdpplot._rs import load_parquet, load_txt, bin_points_py, to_coo
from fastdpplot._rs import parse_fasta_first, parse_fasta_all
from fastdpplot._rs import load_bin_matrix_normalized, load_dp_input

Rust extension availability

If the Rust extension has not been compiled yet, direct imports from fastdpplot._rs will fail with Python's default ModuleNotFoundError.

The high-level wrapper modules such as io and plot provide a friendlier guidance message instead:

fastdpplot._rs is not available. Run 'maturin develop' to build the Rust extension first.

Data contract¶

All rendering functions expect a pandas.DataFrame with three columns:

Column	dtype	Meaning
`x`	`float64`	X coordinate (subject position or arbitrary).
`y`	`float64`	Y coordinate (query position or arbitrary).
`value`	`float32`	Scalar weight (identity, score, or `1.0`).

load() and load_bin() both return this shape.