aristoteleo/dynamo-release

Inclusive model of expression dynamics with conventional or metabolic labeling based scRNA-seq / multiomics, vector field reconstruction and differential geometry analyses

What it solves

Dynamo addresses the limitations of conventional splicing-based RNA velocity analyses in single-cell RNA sequencing (scRNA-seq). It provides a framework to accurately estimate RNA velocities, reconstruct continuous vector fields to predict cell fates, and model time-resolved metabolic labeling to better understand the kinetic regulatory functions governing cell-state transitions.

How it works

The tool uses dynamical systems approaches and differential geometry to map transcriptomic vector fields. It infers absolute RNA velocity and reconstructs continuous vector fields that can identify stable cell types (fixed points) and state boundaries (separatrices). It employs a "least-action-path" method to predict optimal reprogramming paths and uses in silico perturbations to simulate how genetic changes might divert cell fates. It also explicitly models RNA kinetics, including bursting, transcription, splicing, and degradation, specifically for metabolic labeling experiments.

Who it’s for

It is designed for researchers in computational biology and genomics who analyze single-cell multiomics data to study cellular dynamics, cell fate decisions, and regulatory networks.

Highlights

  • Advanced Velocity Estimation: Offers three methods for RNA velocity estimation, including a negative binomial distribution approach and strategies to correct problematic velocity vectors.
  • Inclusive Kinetic Modeling: Supports one-shot, pulse, chase, and mixture metabolic labeling experiments to overcome splicing-based limitations.
  • Mechanistic Insights: Calculates RNA acceleration, curvature, divergence, and the RNA Jacobian to identify master regulators and stability of cell states.
  • Predictive Capabilities: Predicts optimal paths for cell fate reprogramming and the effects of gene-wise perturbations via in silico simulations.

Related

  • Project
    google-deepmind/alphagenome

    A unifying model for deciphering the regulatory code within DNA sequences, providing multimodal predictions for gene expression, splicing, and chromatin features.

  • Project
    scverse/cellrank

    A modular framework for fate mapping in single-cell data that uses Markov state modeling to estimate cellular dynamics and differentiation probabilities.

  • Project
    ai-dynamo/dynamo

    An open-source, datacenter-scale orchestration layer that coordinates multiple inference engines across GPU clusters to maximize throughput and minimize latency.

  • Dispatch
    Arc Virtual Cell Challenge Primer

    The Arc Virtual Cell Challenge tasks participants with training models to predict the effects of gene silencing via CRISPR on cell transcriptomes, utilizing a dataset of 300k single-cell RNA sequencing profiles.

  • Project
    OmicsML/dance

    A deep learning toolkit for single-cell gene expression analysis that provides a benchmark platform for reproducing experiments and automated preprocessing recommendations.