openTSNE: a modular Python library for t-SNE dimensionality reduction and embedding

By Pavlin Gregor Policar, Martin Stražar, Blaz Zupan. 2019

231 citations recorded by the source corpus.

Summary Point-based visualisations of large, multi-dimensional data from molecular biology can reveal meaningful clusters. One of the most popular techniques to construct such visualisations is t-distributed stochastic neighbor embedding (t-SNE), for which a number of extensions have recently been proposed to address issues of scalability and the quality of the resulting visualisations. We introduce openTSNE, a modular Python library that implements the core t-SNE algorithm and its extensions. The library is orders of magnitude faster than existing popular implementations, including those from scikit-learn. Unique to openTSNE is also the mapping of new data to existing embeddings, which can surprisingly assist in solving batch effects. Availability openTSNE is available at https://github.com/pavlin-policar/openTSNE. Contact pavlin.policar@fri.uni-lj.si, blaz.zupan@fri.uni-lj.si

Subfields: Molecular Biology, Representation & Self-Supervised Learning

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Molecular Biology

In Representation & Self-Supervised Learning

Citations in the corpus

Parametric UMAP Embeddings for Representation and Semisupervised Learning (2020, background): “…in two or three dimensions (Van Der Maaten, 2014), and while FIt-SNE can in principle scale to higher dimensions (Linderman et al., 2019), embedding in more than two dimensions is unsupported in both the official implementation (KlugerLab, 2020) and openTSNE (Poliˇcar et al., 2019) U embeddings.”
Embedding to reference t-SNE space addresses batch effects in single-cell classification (2019, methodology): “All experiments were run using openTSNE 4, our open and extensible t-SNE library for Python.”
Deep ViT Features as Dense Visual Descriptors (2021, methodology): “To further contrast the two ViTs, we employ t-SNE [41] to the keys of the last layer [ k(11) i ] , extracted from 50 animal images from PASCAL-Parts [6].”
A Unifying Perspective on Neighbor Embeddings along the Attraction-Repulsion Spectrum (2020, methodology): “4 [26], a Python reimplementation of FIt-SNE [19].”
NOMAD Projection (2025, unknown): “In this map, PubMed documents are vectorized with a custom BERT model, and then reduced to 2 dimensions using OpenTSNE [19], a multi-core CPU implementation of FIt-SNE [12].”