Skip to content

README.md

Latest commit

 

History

History
105 lines (71 loc) · 4.06 KB

README.md

File metadata and controls

105 lines (71 loc) · 4.06 KB

Latest Release DOI

Warp?

Warp is a set of tools for cryo-EM and cryo-ET data processing including, among other tools: Warp, M, WarpTools, MTools, MCore, and Noise2Map.

Citing Warp

https://doi.org/10.5281/zenodo.13982246

This DOI represents all versions and will resolve to the latest version.

Install Warp

Windows

If you want to use Warp on Windows, tutorials and binaries (currently only for v1) can be found at http://www.warpem.com.

Linux

If you're installing from scratch and don't have an environment yet, here is the easiest way to get everything inside a new environment called warp:

conda create -n warp warp -c warpem -c nvidia/label/cuda-12.9.0 -c conda-forge
conda activate warp  # Activate the environment whenever you want to use Warp

If you want to install in an already existing environment:

conda install warp -c warpem -c nvidia/label/cuda-12.9.0 -c conda-forge

If you want to update to the latest version and already have all channels set up in your environment:

conda update warp

Upgrading from v2.0.0dev37 or earlier

Versions up to v2.0.0dev37 used CUDA 11.8 and .NET 8. Starting with v2.0.0dev38, Warp requires CUDA 12.9 and .NET 10. These changes are too large for conda update to handle, so a fresh environment is needed:

conda env remove -n warp
conda create -n warp warp -c warpem -c nvidia/label/cuda-12.9.0 -c conda-forge

You will also need an NVIDIA driver that supports CUDA 12.x (version >= 525.60.13). Check with nvidia-smi.

Use Warp

For information on how to use Warp, M and friends please check out the user guide section of warpem.github.io/warp.

Build Warp on Linux

After cloning this repository, run these commands:

conda env create -f warp_build.yml
conda activate warp_build
./scripts/build-native-unix.sh
./scripts/publish-unix.sh

If you're building on a machine without an NVIDIA GPU, set CONDA_OVERRIDE_CUDA=12.9 before creating the environment so that conda installs the CUDA variant of PyTorch:

CONDA_OVERRIDE_CUDA=12.9 conda env create -f warp_build.yml

All binaries will be in Release/linux-x64/publish.

Here is some inspiration for an lmod module file:

local root = "/path/to/warp/Release/linux-x64/publish"

conflict("warp")

prepend_path("PATH", root)
prepend_path("LD_LIBRARY_PATH", "/path/to/conda_envs/warp_build/lib")
setenv("RELION_EXTERNAL_RECONSTRUCT_EXECUTABLE", pathJoin(root, "Noise2Half"))

Other programs you'll want to install (on Linux)

AreTomo2 seems broken (c.f. #159 and AreTomo2/#21) and we have not added compatibility for AreTomo3 (#279).

We only ensure compatibility with RELION 5.

Editing Documentation

Install mkdocs-material into your conda environment then run

mkdocs serve

To preview the site. This includes hot reloading so you can preview any changes you make.

The documentation is built and deployed by calling mkdocs build on GitHub actions.

Authorship

Warp was originally developed by Dimitry Tegunov in Patrick Cramer's lab at the Max Planck Institute for Biophysical Chemistry in Göttingen, Germany. This code is available in its original repository.

Warp is now being developed by Dimitry Tegunov and Alister Burt at Genentech, Inc. in South San Francisco, USA. For a list of changes that occurred between the last release under the Max Planck Society and the first release at Genentech, please see CHANGELOG.