DISI - User contributions [en]

Creating the IFP environment on wynton

2026-03-23T15:46:17Z

Bwhall61: Created page with "Follow the instructions at: https://github.com/bwhall61/wynton_ifp_scripts"

Follow the instructions at: https://github.com/bwhall61/wynton_ifp_scripts

ViewBoltz

2025-11-19T22:29:34Z

Bwhall61:

Hello boltz'ers,

Have you run boltz-2 on many small molecules?

Are you tired of trying to load, align, and flip through each output cif file?

Do you wish Chimera had a ViewDock utility for easily scrolling through many boltz-2 predictions?

If you answered yes to any of the above then this new utility may be right for you. Claude code has been gracious enough to help me make a ViewDock utility for boltz predictions called ViewBoltz.

== Getting Chimera with ViewBoltz ==

Download and extract the version of Chimera with ViewBoltz from gimel/epyc. It is at <code>/mnt/nfs/soft/chimera/Chimera_w_ViewBoltz.tar.gz</code>. Once extracted, you can run it from the chimera folder with <code>./bin/chimera</code>.

ViewBoltz can be found under Tools -> Surface/Binding Analysis (just like ViewDock)

== Input to ViewBoltz ==

Just provide the parent directory which contains all of the subdirectories containing the boltz-2 output cif files. For example,

<syntaxhighlight lang="text">
parent_folder/
├── mol0/
│ ├── msa/
│ ├── processed/
│ └── predictions/
│ └── 0/
│ ├── 0_model_0.cif
│ └── affinity_0.json
├── mol1/
│ ├── msa/
│ ├── processed/
│ └── predictions/
│ └── 1/
│ ├── 1_model_0.cif
│ └── affinity_1.json
</syntaxhighlight>

In this case you would just provide <code>parent_folder</code>. It doesn't have to be formatted exactly like this, as long as there is some subdirectory that contains the *.cif and affinity_*.json file with the corresponding probability and affinity prediction.

== What should happen ==

Chimera should load each of the predictions into a ViewDock session like you are used to. All of the structures should be aligned. The columns in ViewDock should show the file path for each loaded molecule and the affinity/probability predictions. I think this session should be able to be used like any other ViewDock session.

== Some notes ==

This is super new, was scraped together in an afternoon, and something is bound to break. Please let Brendan know if something isn't working and he can look into it.

H-bonds don't work. Boltz doesn't have a sense of hydrogens so you can't visualize h-bonds like you can in a normal ViewDock session.

ViewBoltz

2025-11-19T22:27:56Z

Bwhall61: Created page with "Hello boltz'ers, Have you run boltz-2 on many small molecules? Are you tired of trying to load, align, and flip through each output cif file? Do you wish Chimera had a ViewDock utility for easily scrolling through many boltz-2 predictions? If you answered yes to any of the above then this new utility may be right for you. Claude code has been gracious enough to help me make a ViewDock utility for boltz predictions called ViewBoltz. == Getting Chimera with ViewBol..."

Hello boltz'ers,

Have you run boltz-2 on many small molecules?

Are you tired of trying to load, align, and flip through each output cif file?

Do you wish Chimera had a ViewDock utility for easily scrolling through many boltz-2 predictions?

If you answered yes to any of the above then this new utility may be right for you. Claude code has been gracious enough to help me make a ViewDock utility for boltz predictions called ViewBoltz.

== Getting Chimera with ViewBoltz ==

Download and extract the version of Chimera with ViewBoltz from gimel/epyc. It is at <code>/mnt/nfs/soft/chimera/Chimera_w_ViewBoltz.tar.gz</code>. Once extracted, you can run it from the chimera folder with `./bin/chimera`.

ViewBoltz can be found under Tools -> Surface/Binding Analysis (just like ViewDock)

== Input to ViewBoltz ==

Just provide the parent directory which contains all of the subdirectories containing the boltz-2 output cif files. For example,

<syntaxhighlight lang="text">
parent_folder/
├── mol0/
│ ├── msa/
│ ├── processed/
│ └── predictions/
│ └── 0/
│ ├── 0_model_0.cif
│ └── affinity_0.json
├── mol1/
│ ├── msa/
│ ├── processed/
│ └── predictions/
│ └── 1/
│ ├── 1_model_0.cif
│ └── affinity_1.json
</syntaxhighlight>

In this case you would just provide <code>parent_folder</code>. It doesn't have to be formatted exactly like this, as long as there is some subdirectory that contains the *.cif and affinity_*.json file with the corresponding probability and affinity prediction.

== What should happen ==

Chimera should load each of the predictions into a ViewDock session like you are used to. All of the structures should be aligned. The columns in ViewDock should show the file path for each loaded molecule and the affinity/probability predictions. I think this session should be able to be used like any other ViewDock session.

== Some notes ==

This is super new, was scraped together in an afternoon, and something is bound to break. Please let Brendan know if something isn't working and he can look into it.

H-bonds don't work. Boltz doesn't have a sense of hydrogens so you can't visualize h-bonds like you can in a normal ViewDock session.

Creating the building docker image

2025-11-18T00:45:31Z

Bwhall61:

= Building Pipeline Docker & Apptainer Images =

Brendan created a Dockerfile for the building pipeline to simplify the flow of submission and reduce the number of intermediate files created, to speed up building when carried out on Beegfs.

This page describes how to create the building pipeline Docker image (on gimel/epyc/etc) and the Apptainer image (on Wynton).

---

== A note on open source ==

Beginning ~Dec 12, 2025 we are beginning to transition away from ChemAxon and OpenEye commercial licenses to use open source tools for our building pipeline. The following instructions will work for both commercial and open source versions. If you have access to the commercial tools, build the docker container with <code>--target full</code> (you can still run the open source software with this image, it will just default to commercial software). If you do not have the commercial tools, build the docker container with <code>--target oss</code>.

== On epyc ==

# Clone the version of the [https://github.com/docking-org/zinc22-3d zinc22-3d] building pipeline you want. For the official version corresponding with the current dock3-release, follow the zinc22-3d link from the [https://github.com/docking-org/dock3-release dock3-release GitHub].
# Run:

<syntaxhighlight lang="bash">
cd zinc22-3d
docker build -t [image_name] \
--target [oss/full] \
--build-context base=[base_dir] \
--build-arg OPENBABEL_VERSION_NUMBER=[version_number] \
--build-context jchem_license_folder=[jchem_license_folder] \
--build-context openeye_license_folder=[openeye_folder] .
</syntaxhighlight>

=== Arguments ===
* '''[oss/full]'''
** oss for open source version
** full for commercial version

* '''image_name'''
** The name of the docker image to make.
** If you are creating the image for everyone in the lab to use, name this ''building_pipeline''.
** For testing, use a different name.

* '''base_dir'''
This is the folder that contains:
<pre>
/soft/jchem-latest
/soft/dock-latest
/soft/corina-latest
/soft/openbabel-latest
/soft/extralibs-latest
/soft/pyenv-latest
</pre>

* '''version_number'''
The version of OpenBabel referenced in:
<pre>
/soft/openbabel-latest/lib/openbabel/[version_number]
/soft/openbabel-latest/share/openbabel/[version_number]
</pre>

* '''jchem_license_folder'''
The folder that contains the JChem license file (''license.cxl''). (Only required for commercial building)

* '''openeye_folder'''
The folder that contains the OpenEye license file (''oe_license.txt''). (Only required for commercial building)

---

=== Example (for lab-wide use only) ===

<syntaxhighlight lang="bash">
docker build -t building_pipeline \
--target full \
--build-context base=/mnt/nfs/soft/dock/versions/dock385/pipeline_3D_ligands \
--build-arg OPENBABEL_VERSION_NUMBER=2.3.1 \
--build-context jchem_license_folder=/nfs/soft/jchem \
--build-context openeye_license_folder=/nfs/soft/openeye .
</syntaxhighlight>

'''Note:''' Only use the name ''building_pipeline'' if you are creating the image for everyone. For testing, please use a different name.

---

== Deploying the Docker Image ==

This will create the building pipeline image on the node you built it on. You now need to load it onto all compute nodes. Run this from an NFS directory (to make sure the logs are saved correctly):

<syntaxhighlight lang="bash">
/mnt/nfs/exk/work/bwhall61/deploy_building_pipeline_docker/deploy_image.sh [image_name]
</syntaxhighlight>

This will queue jobs to load the docker image on all of the compute nodes.

---

== Transferring to Wynton ==

# Save it as a ''.tar'' file:
<syntaxhighlight lang="bash">
docker save -o [image_name].tar [image_name]
</syntaxhighlight>

# Transfer the ''.tar'' file to Wynton.

# Build an Apptainer ''.sif'' image from it:
<syntaxhighlight lang="bash">
apptainer build [image_name].sif docker-archive://[image_name].tar
</syntaxhighlight>

This will give you an Apptainer ''.sif'' file which can then be used for the building pipeline.

Creating the building docker image

2025-09-05T22:01:34Z

Bwhall61: Created page with "= Building Pipeline Docker & Apptainer Images = Brendan created a Dockerfile for the building pipeline to simplify the flow of submission and reduce the number of intermediate files created, to speed up building when carried out on Beegfs. This page describes how to create the building pipeline Docker image (on gimel/epyc/etc) and the Apptainer image (on Wynton). --- == On epyc == # Clone the version of the [https://github.com/docking-org/zinc22-3d zinc22-3d] buil..."

= Building Pipeline Docker & Apptainer Images =

Brendan created a Dockerfile for the building pipeline to simplify the flow of submission and reduce the number of intermediate files created, to speed up building when carried out on Beegfs.

This page describes how to create the building pipeline Docker image (on gimel/epyc/etc) and the Apptainer image (on Wynton).

---

== On epyc ==

# Clone the version of the [https://github.com/docking-org/zinc22-3d zinc22-3d] building pipeline you want. For the official version corresponding with the current dock3-release, follow the zinc22-3d link from the [https://github.com/docking-org/dock3-release dock3-release GitHub].
# Run:

<syntaxhighlight lang="bash">
cd zinc22-3d
docker build -t [image_name] \
--build-context base=[base_dir] \
--build-arg OPENBABEL_VERSION_NUMBER=[version_number] \
--build-context jchem_license_folder=[jchem_license_folder] \
--build-context openeye_license_folder=[openeye_folder] .
</syntaxhighlight>

=== Arguments ===

* '''image_name'''
** The name of the docker image to make.
** If you are creating the image for everyone in the lab to use, name this ''building_pipeline''.
** For testing, use a different name.

* '''base_dir'''
This is the folder that contains:
<pre>
/soft/jchem-latest
/soft/dock-latest
/soft/corina-latest
/soft/openbabel-latest
/soft/extralibs-latest
/soft/pyenv-latest
</pre>

* '''version_number'''
The version of OpenBabel referenced in:
<pre>
/soft/openbabel-latest/lib/openbabel/[version_number]
/soft/openbabel-latest/share/openbabel/[version_number]
</pre>

* '''jchem_license_folder'''
The folder that contains the JChem license file (''license.cxl'').

* '''openeye_folder'''
The folder that contains the OpenEye license file (''oe_license.txt'').

---

=== Example (for lab-wide use only) ===

<syntaxhighlight lang="bash">
docker build -t building_pipeline \
--build-context base=/mnt/nfs/soft/dock/versions/dock385/pipeline_3D_ligands \
--build-arg OPENBABEL_VERSION_NUMBER=2.3.1 \
--build-context jchem_license_folder=/nfs/soft/jchem \
--build-context openeye_license_folder=/nfs/soft/openeye .
</syntaxhighlight>

'''Note:''' Only use the name ''building_pipeline'' if you are creating the image for everyone. For testing, please use a different name.

---

== Deploying the Docker Image ==

This will create the building pipeline image on the node you built it on. You now need to load it onto all compute nodes. Run this from an NFS directory (to make sure the logs are saved correctly):

<syntaxhighlight lang="bash">
/mnt/nfs/exk/work/bwhall61/deploy_building_pipeline_docker/deploy_image.sh [image_name]
</syntaxhighlight>

This will queue jobs to load the docker image on all of the compute nodes.

---

== Transferring to Wynton ==

# Save it as a ''.tar'' file:
<syntaxhighlight lang="bash">
docker save -o [image_name].tar [image_name]
</syntaxhighlight>

# Transfer the ''.tar'' file to Wynton.

# Build an Apptainer ''.sif'' image from it:
<syntaxhighlight lang="bash">
apptainer build [image_name].sif docker-archive://[image_name].tar
</syntaxhighlight>

This will give you an Apptainer ''.sif'' file which can then be used for the building pipeline.