Home

PhosphoVelocity is a Bayesian phosphosite velocity modeling pipeline for LC-MS/MS time-course phosphoproteomics data.

Overview

The pipeline reconstructs latent temporal trajectories of phosphosite intensities from sparse time-course measurements (e.g., triplicates at a handful of time points) using Gaussian Processes, then estimates phosphorylation velocity — the time derivative of the log2 intensity trajectory — as a biologically interpretable signal of kinase activity dynamics.

Key Features

• MaxQuant integration — Parses Phospho (STY)Sites.txt directly.
• GP-based trajectory reconstruction — Scikit-learn GP with Matérn 5/2 + WhiteKernel.
• Bayesian inference — PyMC-based posterior distributions over trajectories and velocities.
• Kinase–substrate network priors — Incorporates network knowledge (KinomeXplorer / NetworKIN).
• CLI — Simple command-line interface for end-to-end runs.

Quick Start

pip install phospho-velocity
phospho-velocity run --input Phospho_STY_Sites.txt --output-dir results/

Citation

If you use phospho-velocity, please cite the relevant methodological papers:

• Robin et al. (2019) — Peptide log-ratio uncertainty model.
• Creixell et al. (2015) — Kinase–substrate network inference.