Skip to content
This repository has been archived by the owner on Sep 19, 2024. It is now read-only.

Commit

Permalink
Deployed a04b4ad with MkDocs version: 1.6.0
Browse files Browse the repository at this point in the history
  • Loading branch information
@jkobject
jkobject committed Aug 7, 2024
1 parent 138f0d3 commit 517f2d0
Show file tree
Hide file tree
Showing 5 changed files with 249 additions and 84 deletions.
Binary file added figure1.png
View file
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
169 changes: 124 additions & 45 deletions index.html
View file
Original file line number Diff line number Diff line change
Expand Up @@ -43,16 +43,40 @@
<ul class="current">
<li class="toctree-l1 current"><a class="reference internal current" href="#">Home</a>
<ul class="current">
<li class="toctree-l2"><a class="reference internal" href="#install-it-from-pypi">Install it from PyPI</a>
<li class="toctree-l2"><a class="reference internal" href="#install-scprint">Install scPRINT</a>
<ul>
<li class="toctree-l3"><a class="reference internal" href="#laminai">lamin.ai</a>
</li>
<li class="toctree-l2"><a class="reference internal" href="#usage">Usage</a>
</ul>
</li>
<li class="toctree-l2"><a class="reference internal" href="#development">Development</a>
<li class="toctree-l2"><a class="reference internal" href="#usage">Usage</a>
<ul>
<li class="toctree-l3"><a class="reference internal" href="#what-is-included">What is included?</a>
<li class="toctree-l3"><a class="reference internal" href="#scprints-basic-commands">scPRINT's basic commands</a>
</li>
<li class="toctree-l3"><a class="reference internal" href="#notes-on-gpucpu-usage-with-triton">Notes on GPU/CPU usage with triton</a>
</li>
<li class="toctree-l3"><a class="reference internal" href="#i-want-to-generate-gene-networks-from-scrnaseq-data">I want to generate gene networks from scRNAseq data:</a>
</li>
<li class="toctree-l3"><a class="reference internal" href="#i-want-to-generate-cell-embeddings-and-cell-label-predictions-from-scrnaseq-data">I want to generate cell embeddings and cell label predictions from scRNAseq data:</a>
</li>
<li class="toctree-l3"><a class="reference internal" href="#i-want-to-denoising-my-scrnaseq-dataset">I want to denoising my scRNAseq dataset:</a>
</li>
<li class="toctree-l3"><a class="reference internal" href="#i-want-to-generate-an-atlas-level-embedding">I want to generate an atlas-level embedding</a>
</li>
<li class="toctree-l3"><a class="reference internal" href="#i-need-to-generate-gene-tokens-using-pllms">I need to generate gene tokens using pLLMs</a>
</li>
<li class="toctree-l3"><a class="reference internal" href="#i-want-to-pre-train-scprint-from-scratch-on-my-own-data">I want to pre-train scPRINT from scratch on my own data</a>
</li>
<li class="toctree-l3"><a class="reference internal" href="#documentation">Documentation</a>
</li>
<li class="toctree-l3"><a class="reference internal" href="#model-weights">Model Weights</a>
</li>
</ul>
</li>
<li class="toctree-l2"><a class="reference internal" href="#development">Development</a>
</li>
<li class="toctree-l2"><a class="reference internal" href="#work-in-progress">Work in progress:</a>
</li>
</ul>
</li>
</ul>
Expand Down Expand Up @@ -111,60 +135,115 @@
<div role="main" class="document" itemscope="itemscope" itemtype="http://schema.org/Article">
<div class="section" itemprop="articleBody">

<h1 id="scprint">scprint</h1>
<p><a href="https://codecov.io/gh/jkobject/scPRINT"><img alt="codecov" src="https://codecov.io/gh/jkobject/scPRINT/branch/main/graph/badge.svg?token=scPRINT_token_here" /></a>
<a href="https://github.com/jkobject/scPRINT/actions/workflows/main.yml"><img alt="CI" src="https://github.com/jkobject/scPRINT/actions/workflows/main.yml/badge.svg" /></a></p>
<p>Awesome Large Transcriptional Model created by Jeremie Kalfon</p>
<p>scprint = single cell pretrained regulation inference neural network from transcripts</p>
<p>using: </p>
<h2 id="install-it-from-pypi">Install it from PyPI</h2>
<p>first have a good version of pytorch installed</p>
<p>you might need to make it match your cuda version etc..</p>
<p>We only support torch&gt;=2.0.0</p>
<p>then install laminDB</p>
<pre><code class="language-bash">pip install 'lamindb[jupyter,bionty]'
</code></pre>
<p>then install scPrint</p>
<pre><code class="language-bash">pip install scprint

I had to install a specific version of pytorch, torchaudio, torchtext.. for my cuda version.
My cuda compiler nvcc compiles cuda 11.7. my cuda-smi (api) is currently 12.1.

Please install all of it for your cuda version and it should still work.

for more information on this, please see [installation.md](installation.md).
<h1 id="scprint-large-cell-model-for-scrnaseq-data">scPRINT: Large Cell Model for scRNAseq data</h1>
<p><a href="https://badge.fury.io/py/scprint"><img alt="PyPI version" src="https://badge.fury.io/py/scprint.svg" /></a>
<a href="https://scprint.readthedocs.io/en/latest/?badge=latest"><img alt="Documentation Status" src="https://readthedocs.org/projects/scprint/badge/?version=latest" /></a>
<a href="https://pepy.tech/project/scprint"><img alt="Downloads" src="https://pepy.tech/badge/scprint" /></a>
<a href="https://pepy.tech/project/scprint"><img alt="Downloads" src="https://pepy.tech/badge/scprint/month" /></a>
<a href="https://pepy.tech/project/scprint"><img alt="Downloads" src="https://pepy.tech/badge/scprint/week" /></a>
<a href="https://img.shields.io/github/issues/jkobject/scPRINT"><img alt="GitHub issues" src="https://img.shields.io/github/issues/jkobject/scPRINT" /></a>
<a href="https://github.com/psf/black"><img alt="Code style: black" src="https://img.shields.io/badge/code%20style-black-000000.svg" /></a>
<a href=""><img alt="DOI" src="https://zenodo.org/badge/391909874.svg" /></a></p>
<p><img alt="logo" src="logo.png" /></p>
<p>scPRINT is a large transformer model built for the inference of gene networks (connections between genes explaining the cell's expression profile) from scRNAseq data.</p>
<p>It uses novel encoding and decoding of the cell expression profile and new pre-training methodologies to learn a cell model.</p>
<p>scPRINT can be used to perform the following analyses:</p>
<ul>
<li><strong>expression denoising</strong>: increase the resolution of your scRNAseq data</li>
<li><strong>cell embedding</strong>: generate a low-dimensional representation of your dataset</li>
<li><strong>label prediction</strong>: predict the cell type, disease, sequencer, sex, and ethnicity of your cells</li>
<li><strong>gene network inference</strong>: generate a gene network from any cell or cell cluster in your scRNAseq dataset</li>
</ul>
<p><a href="https://www.biorxiv.org/content/10.1101/2024.07.29.605556v1">Read the paper!</a> if you would like to know more about scPRINT.</p>
<p><img alt="figure1" src="figure1.png" /></p>
<h2 id="install-scprint">Install <code>scPRINT</code></h2>
<p>For the moment scPRINT has been tested on MacOS and Linux (Ubuntu 20.04) with Python 3.10.</p>
<p>If you want to be using flashattention2, know that it only supports triton 2.0 MLIR's version and torch==2.0.0 for now.</p>
<pre><code class="language-python">conda create -n &quot;[whatever]&quot; python==3.10
git clone https://github.com/jkobject/scPRINT
#one of
pip install scPRINT # OR
pip install scPRINT[dev] # for the dev dependencies (building etc..) AND/OR [dev,flash]
pip install scPRINT[flash] &amp;&amp; pip install -e &quot;git+https:/
/github.com/triton-lang/triton.git@legacy-backend
#egg=triton&amp;subdirectory=python&quot; # to use flashattention2, you will need to install triton 2.0.0.dev20221202 specifically, working on removing this dependency # only if you have a compatible gpu (e.g. not available for apple GPUs for now, see https://github.com/triton-lang/triton?tab=readme-ov-file#compatibility)
</code></pre>
<p>We make use of some additional packages we developed alongside scPRint.</p>
<p>Please refer to their documentation for more information:</p>
<ul>
<li><a href="https://github.com/jkobject/scDataLoader">scDataLoader</a>: a dataloader for training large cell models.</li>
<li><a href="https://github.com/cantinilab/GRnnData">GRnnData</a>: a package to work with gene networks from single cell data.</li>
<li><a href="https://github.com/jkobject/benGRN">benGRN</a>: a package to benchmark gene network inference methods from single cell data.</li>
</ul>
<h3 id="laminai">lamin.ai</h3>
<p>⚠️ if you want to use the scDataloader's multi-dataset mode or if you want to preprocess datasets and other functions of the model, you will need to use lamin.ai.</p>
<p>In that case, connect with google or github to <a href="https://lamin.ai/login">lamin.ai</a>, then be sure to connect before running anything (or before starting a notebook): <code>lamin login &lt;email&gt; --key &lt;API-key&gt;</code>. Follow the instructions on <a href="https://docs.lamin.ai/guide">their website</a>.</p>
<h2 id="usage">Usage</h2>
<h3 id="scprints-basic-commands">scPRINT's basic commands</h3>
<p>This is the most minimal example of how scPRINT works:</p>
<pre><code class="language-py">from lightning.pytorch import Trainer
from scprint import scPrint
from scdataloader import DataModule

...
datamodule = DataModule(...)
model = scPrint(...)
# to train / fit / test the model
trainer = Trainer(...)
trainer.fit(model, datamodule=datamodule)
# to do predictions Denoiser, Embedder, GNInfer
denoiser = Denoiser(...)
adata = sc.read_h5ad(...)
denoiser(model, adata=adata)
...
</code></pre>
<pre><code class="language-bash">$ python -m scPrint/__main__.py
#or
$ scprint fit/train/predict/test
<p>or, from a bash command line</p>
<pre><code class="language-bash">$ scprint fit/train/predict/test/denoise/embed/gninfer --config config/[medium|large|vlarge] ...
</code></pre>
<p>for more information on usage please see the documentation in https://jkobject.com/scPrint</p>
<p>find out more about the commands by running <code>scprint --help</code> or <code>scprint [command] --help</code>.</p>
<p>more examples of using the command line are available in the <a href="./docs/usage.md">docs</a>.</p>
<h3 id="notes-on-gpucpu-usage-with-triton">Notes on GPU/CPU usage with triton</h3>
<p>If you do not have <a href="https://triton-lang.org/main/python-api/triton.html">triton</a> installed you will not be able to take advantage of GPU acceleration, but you can still use the model on the CPU.</p>
<p>In that case, if loading from a checkpoint that was trained with flashattention, you will need to specify <code>transformer="normal"</code> in the <code>load_from_checkpoint</code> function like so:</p>
<pre><code class="language-python">model = scPrint.load_from_checkpoint(
'../data/temp/last.ckpt', precpt_gene_emb=None,
transformer=&quot;normal&quot;)
</code></pre>
<p>We now explore the different usages of scPRINT:</p>
<h3 id="i-want-to-generate-gene-networks-from-scrnaseq-data">I want to generate gene networks from scRNAseq data:</h3>
<p>-&gt; Refer to the section . gene network inference in <a href="notebooks/cancer_usecase/">this notebook</a>.</p>
<p>-&gt; More examples in this notebook <a href="../notebooks/bench_omni.ipynb">notebooks/assessments/bench_omni.ipynb</a>.</p>
<h3 id="i-want-to-generate-cell-embeddings-and-cell-label-predictions-from-scrnaseq-data">I want to generate cell embeddings and cell label predictions from scRNAseq data:</h3>
<p>-&gt; Refer to the embeddings and cell annotations section in <a href="notebooks/cancer_usecase/">this notebook</a>.</p>
<h3 id="i-want-to-denoising-my-scrnaseq-dataset">I want to denoising my scRNAseq dataset:</h3>
<p>-&gt; Refer to the Denoising of B-cell section in <a href="notebooks/cancer_usecase/">this notebook</a>.</p>
<p>-&gt; More example in our benchmark notebook <a href="../notebooks/bench_denoising.ipynb">notebooks/assessments/bench_denoising.ipynb</a>.</p>
<h3 id="i-want-to-generate-an-atlas-level-embedding">I want to generate an atlas-level embedding</h3>
<p>-&gt; Refer to the notebook <a href="../figures/nice_umap.ipynb">figures/nice_umap.ipynb</a>.</p>
<h3 id="i-need-to-generate-gene-tokens-using-pllms">I need to generate gene tokens using pLLMs</h3>
<p>To run scPRINT, you can use the option to define the gene tokens using protein language model embeddings of genes. This is done by providing the path to a parquet file of the precomputed set of embeddings for each gene name to scPRINT via "precpt_gene_emb"</p>
<p>-&gt; To generate this file please refer to the notebook <a href="../notebooks/generate_gene_embeddings.ipynb">notebooks/generate_gene_embeddings.ipynb</a>.</p>
<h3 id="i-want-to-pre-train-scprint-from-scratch-on-my-own-data">I want to pre-train scPRINT from scratch on my own data</h3>
<p>-&gt; Refer to the documentation page <a href="pretrain/">pretrain scprint</a></p>
<h3 id="documentation">Documentation</h3>
<p>For more information on usage please see the documentation in <a href="https://www.jkobject.com/scPrint/">https://www.jkobject.com/scPrint/</a></p>
<h3 id="model-weights">Model Weights</h3>
<p>Model weights are available on <a href="https://huggingface.co/jkobject/scPRINT/">hugging face</a>.</p>
<h2 id="development">Development</h2>
<p>Read the <a href="CONTRIBUTING.md">CONTRIBUTING.md</a> file.</p>
<h3 id="what-is-included">What is included?</h3>
<ul>
<li>📃 Documentation structure using <a href="http://www.mkdocs.org">mkdocs</a></li>
<li>🧪 Testing structure using <a href="https://docs.pytest.org/en/latest/">pytest</a>
If you want <a href="https://about.codecov.io/sign-up/">codecov</a> Reports and Automatic Release to <a href="https://pypi.org">PyPI</a><br />
On the new repository <code>settings-&gt;secrets</code> add your <code>PYPI_API_TOKEN</code> and <code>CODECOV_TOKEN</code> (get the tokens on respective websites)</li>
<li>✅ Code linting using <a href="https://flake8.pycqa.org/en/latest/">flake8</a></li>
<li>📊 Code coverage reports using <a href="https://about.codecov.io/sign-up/">codecov</a></li>
<li>🛳️ Automatic release to <a href="https://pypi.org">PyPI</a> using <a href="https://twine.readthedocs.io/en/latest/">twine</a> and github actions.</li>
</ul>
<p>acknowledgement:
<p>Read the <a href="https://wandb.ai/ml4ig/scprint_scale/reports/scPRINT-trainings--Vmlldzo4ODIxMjgx?accessToken=80metwx7b08hhourotpskdyaxiflq700xzmzymr6scvkp69agybt79l341tv68hp">training runs</a> document to know more about how pre-training was performed and the its behavior.</p>
<p>Acknowledgement:
<a href="https://github.com/rochacbruno/python-project-template">python template</a>
<a href="">scGPT</a>
<a href="">laminDB</a></p>
<a href="https://lamin.ai/">laminDB</a>
<a href="https://lightning.ai/">lightning</a></p>
<h2 id="work-in-progress">Work in progress:</h2>
<ol>
<li>remove the triton dependencies</li>
<li>add version with additional labels (tissues, age) and organisms (mouse, zebrafish) and more datasets from cellxgene</li>
<li>version with separate transformer blocks for the encoding part of the bottleneck learning and for the cell embeddings</li>
<li>improve classifier to output uncertainties and topK predictions when unsure</li>
<li></li>
</ol>
<p>Awesome Large Cell Model created by Jeremie Kalfon.</p>

</div>
</div><footer>
Expand Down Expand Up @@ -213,5 +292,5 @@ <h3 id="what-is-included">What is included?</h3>

<!--
MkDocs version : 1.6.0
Build Date UTC : 2024-08-07 12:53:32.690372+00:00
Build Date UTC : 2024-08-07 14:26:42.295691+00:00
-->
Loading

0 comments on commit 517f2d0

Please sign in to comment.