Exploring dimension-reduced embeddings

Last update: Nov 29, 2022

Related tags

Text Data & NLP sleepwalk

Overview

sleepwalk

Exploring dimension-reduced embeddings

This is the code repository. See here for the Sleepwalk web page.

License and disclaimer

This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.

This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.

You should have received a copy of the GNU General Public License along with this program. If not, see https://www.gnu.org/licenses/.

Comments

Error running sleepwalk: cannot open the connection
Dear sleepwalk developers, Thanks a lot for providing such nice method. I could install the package but I get the following error when I tried to run:

> sleepwalk([email protected][email protected], [email protected][email protected]) Estimating 'maxdist' for feature matrix 1 Server has been stopped. Server has been stopped. Error in app$openPage(useViewer, browser) : Timeout waiting for websocket. In addition: Warning messages: 1: In file(con, "r") : cannot open file 'sleepwalk_canvas.html': No such file or directory 2: In func(req) : File '/favicon.ico' is not found

I know this is probably not a sleepwalk specific error, but I couldn't find a solution for this. Any hints/help on how to fix this issue?

Also, I have a question about the output. Besides using the interactive mode to manually inspect cells that might be "misplaced" on the reduced-dimension space, I would like to systematically find the cells that don't quite fit to the clusters they were originally assigned to. In other words, how would you suggest to use sleepwalk to refine my clustering since I suspect that many of my cells were wrongly assigned to their clusters. I am using Seurat package to reduce dimension and clustering.

Thank you very much, Gustavo
opened by gufranca 2
Error: 'browser' must be a non-empty character string
Hello,

After calling the sleepwalk function on a Seurat object, I got this error:

> sleepwalk( as.matrix([email protected][email protected]), as.matrix([email protected][email protected]) ) Estimating 'maxdist' for feature matrix 1 Error in browseURL(str_c("http://localhost:", port, "/", pageobj$startPage), : 'browser' must be a non-empty character string

I have loaded the stringr library (containing the function str_c()), and I cannot find the file originating this error. Can I ask if someone had this problem at some point?

Thank you
opened by PedroRaposo 2
slw_on_selection error when sleepwalk is not attached

Running sleepwalk without attaching the package (i.e., NOT specifying library(sleepwalk)) like this works fine:

sleepwalk::sleepwalk(se[email protected][email protected], t([email protected][[email protected],]))

But the moment you select cells with your mouse, it crashed (browser tab closes) and R gives this error:

Error in slw_on_selection(selPoints, 1) : could not find function "slw_on_selection"

Loading the package using library(sleepwalk) solves the issue, but it'd be nice if it weren't necessary.

opened by FelixTheStudent 0
doc for comparison

The example on the web page for comparing two embeddings still uses the old version where both distances are used concurrently. We also need to change the explanation below to say that the same cell always has the same colour in all embeddings

opened by simon-anders 0
Suggestion: Link embeddings from transposed table

Let say I have e.g. a matrix where I have individuals (cells e.g.) as rows and features as columns, and then run a UMAP on both the ordinary matrix, and the transposed one. Then it would be natural to want to look at the individual UMAP with the default usage (the distances to other individuals), but it would also be interesting to see the features for that individual (and vice versa).

Is it clear what I mean?

opened by StaffanBetner 2

Releases(v0.3.2)

v0.3.2(Sep 17, 2021)
jrc now (v.0.5.0) uses setLimits function for all the security restriction. This update fixes the dependency problem caused by that change.

Source code(tar.gz)
Source code(zip)
v0.3.1(Sep 30, 2020)
broken path to the start page, caused by jrc update fixed

Source code(tar.gz)
Source code(zip)
v.0.3.0(Feb 27, 2020)
New argument metric allows to use angular distance (metric = "cosine") as an alternative to default Euclidean distance (meric = "euclid").

If compare = "distances", it is no longer required to provide several embeddings. If only one embedding is given, it will be used for all the distances.

Source code(tar.gz)
Source code(zip)
v0.2.1(Oct 2, 2019)
Changes due to an update of the jrc package.

Indices of selected points are no longer stored in a variable and can be accessed only via the callback function. Thus, no changes to the global environment are made, unless user specifies them his- or herself.

Added the possibility to pass arguments to jrc::openPage (such as port number or browser in which to open the app.)

Source code(tar.gz)
Source code(zip)
v0.2.0(Sep 27, 2019)
Now HTML Canvas is used to plot the embedding. It makes Sleepwalk faster and allows to simultaneously display more points.

New parameter mode = c("canvas", "svg") is added, that allows user to go back to the old SVG-based version of Sleepwalk app.

Bug in slw_snapshot is fixed. The function no longer returns a list of identical plots, when used with several different embeddings.

Source code(tar.gz)
Source code(zip)

Owner

S. Anders's research group at ZMBH

GitHub Repository https://anders-biostat.github.io/sleepwalk/

"Investigating the Limitations of Transformers with Simple Arithmetic Tasks", 2021

transformers-arithmetic This repository contains the code to reproduce the experiments from the paper: Nogueira, Jiang, Lin "Investigating the Limitat

33 Nov 16, 2022

A Non-Autoregressive Transformer based TTS, supporting a family of SOTA transformers with supervised and unsupervised duration modelings. This project grows with the research community, aiming to achieve the ultimate TTS.

A Non-Autoregressive Transformer based TTS, supporting a family of SOTA transformers with supervised and unsupervised duration modelings. This project grows with the research community, aiming to ach

237 Jan 02, 2023

Code repository of the paper Neural circuit policies enabling auditable autonomy published in Nature Machine Intelligence

9 Jan 08, 2023

🧪 Cutting-edge experimental spaCy components and features

spacy-experimental: Cutting-edge experimental spaCy components and features This package includes experimental components and features for spaCy v3.x,

65 Dec 30, 2022

Exploring dimension-reduced embeddings

Related tags

Overview

sleepwalk

License and disclaimer

Comments

Error running sleepwalk: cannot open the connection

Error: 'browser' must be a non-empty character string

slw_on_selection error when sleepwalk is not attached

doc for comparison

Suggestion: Link embeddings from transposed table

Releases(v0.3.2)

v0.3.2(Sep 17, 2021)

v0.3.1(Sep 30, 2020)

v.0.3.0(Feb 27, 2020)

v0.2.1(Oct 2, 2019)

v0.2.0(Sep 27, 2019)

Owner

S. Anders's research group at ZMBH

"Investigating the Limitations of Transformers with Simple Arithmetic Tasks", 2021

A Non-Autoregressive Transformer based TTS, supporting a family of SOTA transformers with supervised and unsupervised duration modelings. This project grows with the research community, aiming to achieve the ultimate TTS.

Using BERT-based models for toxic span detection

End-2-end speech synthesis with recurrent neural networks

A notebook that shows how to import the IITB English-Hindi Parallel Corpus from the HuggingFace datasets repository

BookNLP, a natural language processing pipeline for books

Chinese real time voice cloning (VC) and Chinese text to speech (TTS).

The official implementation of VAENAR-TTS, a VAE based non-autoregressive TTS model.

This is the source code of RPG (Reward-Randomized Policy Gradient)

A collection of GNN-based fake news detection models.

Open-source offline translation library written in Python. Uses OpenNMT for translations

Th2En & Th2Zh: The large-scale datasets for Thai text cross-lingual summarization

NLPIR tutorial: pretrain for IR. pre-train on raw textual corpus, fine-tune on MS MARCO Document Ranking

Unofficial Parallel WaveGAN (+ MelGAN & Multi-band MelGAN & HiFi-GAN & StyleMelGAN) with Pytorch

A script that automatically creates a branch name using google translation api and jira api

topic modeling on unstructured data in Space news articles retrieved from the Guardian (UK) newspaper using API

ADCS cert template modification and ACL enumeration

This is an incredibly powerful calculator that is capable of many useful day-to-day functions.

Code repository of the paper Neural circuit policies enabling auditable autonomy published in Nature Machine Intelligence

🧪 Cutting-edge experimental spaCy components and features