ssciwr · ChristineSchulz · Nov 18, 2025 · Nov 19, 2025 · Dec 1, 2025 · Dec 1, 2025
diff --git a/.github/workflows/main.yml b/.github/workflows/main.yml
@@ -32,16 +32,44 @@ jobs:
     - name: Install dependencies
       run: |
         python -m pip install --upgrade pip
-        pip install -r requirements.txt
-    - name: Run Alignment notebook
+        python -m pip install -r requirements.txt
+    - name: Pre-cache matplotlib font (cross-platform)
+      run: |
+        python -c "import os; os.environ['MPLBACKEND']='Agg'; import matplotlib.pyplot"
+    - name: Run AlphaFold3  Alignment notebook
       run: |
         cd notebooks
         python -m pytest --nbval AFold_Alignment_CPU.ipynb
-    - name: Run Inference notebook
+    - name: Run AlphaFold3 Inference notebook
       run: |
         cd notebooks
         python -m pytest --nbval AFold_Diffusion_GPU.ipynb
-    - name: Run Analysis
+    - name: Run AlphaFold3 Analysis notebook
       run: |  
         cd notebooks
         python -m pytest --nbval AFold_Confidence_Levels.ipynb
+
+    - name: Run Bindcraft notebook
+      run: |  
+        cd notebooks
+        python -m pytest --nbval bindcraft.ipynb
+
+    - name: Run Boltz Inference notebook
+      run: |  
+        cd notebooks
+        python -m pytest --nbval boltz_input.ipynb
+
+    - name: Run Boltz Analysis notebook
+      run: |  
+        cd notebooks
+        python -m pytest --nbval boltz_confidence_levels.ipynb
+
+    - name: Run Boltzgen
+      run: |  
+        cd notebooks
+        python -m pytest --nbval boltzgen.ipynb
+
+    - name: Run RFDiffusion
+      run: |  
+        cd notebooks
+        python -m pytest --nbval RFDiffusion.ipynb
diff --git a/.pre-commit-config.yaml b/.pre-commit-config.yaml
@@ -7,7 +7,7 @@ repos:
       - id: ruff-check
       # Run the formatter.
       - id: ruff-format
-  - repo: https://github.com/kynan/nbstripout
-    rev: 0.8.2
-    hooks:
-    - id: nbstripout
+#  - repo: https://github.com/kynan/nbstripout
+#    rev: 0.8.2
+#    hooks:
+#    - id: nbstripout
diff --git a/docs/about.md b/docs/about.md
@@ -6,12 +6,4 @@ hide:
 # About
 The prediction of biomolecular structures is routed in the translation from protein or nucleic acid sequence to 3D structure. In reality, these 3D structures result from a delicate interplay with small molecules, ions, fatty acids and solvents. At the same time, the predicted structures are a product of the underlying machine-learning models. By combining the applicability range of the method with the limitations of modeling biological systems, we can provide confidence estimates in the context of the respective research question. We aim to extend our offer beyond general models like AlphaFold to more specific tools in the field of sequence based predictions to provide researchers with the ideal tools to their specific needs. 
 
-Embedding the Bio-Structure Hub in the SSC enables, building research software sustainably and in accordance with good scientific practice. This entails on the one hand making use of software engineering tools and methods such as version control, development and production environments, testing frameworks, documentation and release workflows, and a development process, to name a few; and on the other hand, acknowledging that research software is an infrastructure that is the foundation of cutting-edge research, and as such needs to be drafted, designed, operated and maintained in a purposeful manner.
-
-## Projects
-
---8<--
-projects.md:2:3
---8<--
-
-A list of current projects is provided [here](projects.md).
+Embedding the Bio-Structure Hub in the SSC enables, building research software sustainably and in accordance with good scientific practice. This entails on the one hand making use of software engineering tools and methods such as version control, development and production environments, testing frameworks, documentation and release workflows, and a development process, to name a few; and on the other hand, acknowledging that research software is an infrastructure that is the foundation of cutting-edge research, and as such needs to be drafted, designed, operated and maintained in a purposeful manner.
diff --git a/docs/images/favicon.png b/docs/images/favicon.png
diff --git a/docs/images/tutorial/bwVisu_Bindcraft_files.png b/docs/images/tutorial/bwVisu_Bindcraft_files.png
diff --git a/docs/images/tutorial/bwVisu_Bindcraft_input.png b/docs/images/tutorial/bwVisu_Bindcraft_input.png
diff --git a/docs/images/tutorial/bwVisu_Bindcraft_modules_list.png b/docs/images/tutorial/bwVisu_Bindcraft_modules_list.png
diff --git a/docs/images/tutorial/bwVisu_Bindcraft_modules_loaded.png b/docs/images/tutorial/bwVisu_Bindcraft_modules_loaded.png
diff --git a/docs/images/tutorial/bwVisu_Bindcraft_output.png b/docs/images/tutorial/bwVisu_Bindcraft_output.png
diff --git a/docs/images/tutorial/bwVisu_Boltz_MSA.png b/docs/images/tutorial/bwVisu_Boltz_MSA.png
diff --git a/docs/images/tutorial/bwVisu_Boltz_done.png b/docs/images/tutorial/bwVisu_Boltz_done.png
diff --git a/docs/images/tutorial/bwVisu_Boltz_output.png b/docs/images/tutorial/bwVisu_Boltz_output.png
diff --git a/docs/images/tutorial/bwVisu_Boltzgen_files.png b/docs/images/tutorial/bwVisu_Boltzgen_files.png
diff --git a/docs/images/tutorial/bwVisu_GPU_Kernel.png b/docs/images/tutorial/bwVisu_GPU_Kernel.png
diff --git a/docs/images/tutorial/bwVisu_RFD_files.png b/docs/images/tutorial/bwVisu_RFD_files.png
diff --git a/docs/images/tutorial/restart_kernel.png b/docs/images/tutorial/restart_kernel.png
diff --git a/docs/projects.md b/docs/projects.md
@@ -1,2 +1,7 @@
+---
+hide:
+ - navigation
+---
+
 # Projects
 Current projects in the Bio-Structure Hub range from carefully cofolding the components of large protein complex structures, to adding molecular cofactors to improve the quality of predicted structures, or modeling interaction sites for various species. We assist in leveraging structure predictions to plan future experiments, or run preliminary simulations to be used in proposals for future projects.
diff --git a/docs/resources.md → docs/resources/learning.md b/docs/resources.md → docs/resources/learning.md
@@ -1,31 +1,12 @@
+# Learning Center - Training Resources
 
-
-## Methods and Access
-
-We offer a range of resources from general structure prediction methods to specialized tools. These can be accessed on your local computer, using cloud solutions, or in high-performance computing environments.
-
-### General Structure Prediction Methods
-
-We offer support in using and interpreting the results of general models such as [AlphaFold3](https://doi.org/10.1038/s41586-024-07487-w) or [Boltz-1](https://doi.org/10.1101/2024.11.19.624167). For specialized tasks beyond the capabilities of general methods we offer custom solutions, to find the best method for your project!
-
-
-### Computational Infrastructure
-
-The hub provides essential support for accessing structure prediction tools via both High Performance Computing (HPC) and Cloud Solutions:
-
-For HPC access,  AlphaFold 2 and AlphaFold 3 are available on the [bwForCluster Helix](https://wiki.bwhpc.de/e/Helix). This includes interactive access via the [bwVisu service](https://wiki.bwhpc.de/e/Helix/bwVisu). We offer assistance and training for cluster usage.
-
-Regarding cloud computing, we offer support and share best practices around the existing cloud computing options, such as the [AlphaFold3 server]( https://alphafoldserver.com/ ) or the [ProteinAI service](https://protein-ai.academiccloud.de/) for Boltz-1 and AlphaFold2.
-
-## Learning Center - Training Resources
-
-### Introduction to Bioinformatics
+## Introduction to Bioinformatics
 
 Are you new to the world of bioinformatics and not sure where to start? Don't worry, we've got you covered! The EMBL-EBI offers an excellent resource for beginners with their online course ["Bioinformatics for the terrified"](https://www.ebi.ac.uk/training/online/courses/bioinformatics-terrified/). This course provides broad overview of how computers are used in biology, covering the basics and beyond. It's the perfect starting point for anyone looking to dip their toes into the world of bioinformatics. 
 
 A great resource for virtual lectures and webinars is [TESS by the elixir network](https://tess.elixir-europe.org/materials). These free resources offer a great start into any topic in bioinformatics.
 
-### 3D Structure Prediction
+## 3D Structure Prediction
 For a light introduction to protein structure prediction, an overview on the history of AlphaFold and its first appearance in the CASP competition, and more, we recommend checking out this [video by Veritasium on Youtube](https://www.youtube.com/watch?v=P_fHJIYENdI). 
 
 Among other courses in the area of bioinformatics,  the EMBL-EBI also offers more specialized training with their course [Alphafold2](https://www.ebi.ac.uk/training/online/courses/alphafold/). This course goes into the practical aspects of structure predictions, providing hands-on experience and expert guidance. If you prefer video formats, there is a great [talk by Simon Kohl](https://www.youtube.com/watch?v=tTN0MM2CQLU) during the heidelberg.ai series on Youtube talking about the model and concepts behind AlphaFold 2.

diff --git a/docs/resources/methods.md b/docs/resources/methods.md
@@ -0,0 +1,16 @@
+# Methods and Access
+
+We offer a range of resources from general structure prediction methods to specialized tools. These can be accessed on your local computer, using cloud solutions, or in high-performance computing environments.
+
+## General Structure Prediction Methods
+
+We offer support in using and interpreting the results of general models such as [AlphaFold3](https://doi.org/10.1038/s41586-024-07487-w) or [Boltz-1](https://doi.org/10.1101/2024.11.19.624167). For specialized tasks beyond the capabilities of general methods we offer custom solutions, to find the best method for your project!
+
+
+## Computational Infrastructure
+
+The hub provides essential support for accessing structure prediction tools via both High Performance Computing (HPC) and Cloud Solutions:
+
+For HPC access,  AlphaFold 2 and AlphaFold 3 are available on the [bwForCluster Helix](https://wiki.bwhpc.de/e/Helix). This includes interactive access via the [bwVisu service](https://wiki.bwhpc.de/e/Helix/bwVisu). We offer assistance and training for cluster usage.
+
+Regarding cloud computing, we offer support and share best practices around the existing cloud computing options, such as the [AlphaFold3 server]( https://alphafoldserver.com/ ) or the [ProteinAI service](https://protein-ai.academiccloud.de/) for Boltz-1 and AlphaFold2.
diff --git a/docs/services/collab.md b/docs/services/collab.md
@@ -0,0 +1,9 @@
+# Collaboration
+Recurring support requests in the same project or complex tasks that require dedicated research can lead to a longer-term collaboration. We support third-party funding applications, which can secure dedicated time for a project.
+
+# Training
+
+For larger projects we also offer a unique collaboration model by training a member of your team to work independently on the project. Our training program includes continuous support and access to resources for independent work. This includes advice on how to find and use appropriate resources such as documentation, software and pipelines. This process is tailored to the individual question at hand.
+
+
+[Contact us!](mailto:ssc-biostructurehub@uni-heidelberg.de)
diff --git a/docs/services.md → docs/services/service.md b/docs/services.md → docs/services/service.md
@@ -1,3 +1,4 @@
+# Service
 
 We provide software development and scientific support with structure predictions of biomolecules.
 Get in touch with us to see if we can help you!
@@ -15,16 +16,4 @@ The first step is to get in touch with us - our services are free of charge. Whe
 The initial consultation takes about one hour. In this initial consultation, we clarify further aspects and then suggest an approach moving forward: Either providing you with resources, or investigating your question for you in a small-scale project.
 
 ## Project support
-Following up on a consultation, projects can receive support. This could be a small development script, feedback to existing software, or entire prediction projects. 
-
-## Collaboration
-Recurring support requests in the same project can lead to a longer-term collaboration. We support third-party funding applications.
-
-## Training
-
-For larger projects we also offer collaborations by training a member of your team. We offer continuous support and resources for independent work. This includes advice on how to find and use appropriate resources such as documentation, software and pipelines.
-
-
-## Teaching
-Best practices and guides will be collected in tutorials and (virtual) coursework.
-This includes applications and support in providing routes to access and use compute resources for bio-structure predictions.
+Following up on a consultation, projects can receive support. This could be a small development script, feedback to existing software, or entire prediction projects. 
diff --git a/docs/services/teaching.md b/docs/services/teaching.md
@@ -0,0 +1,8 @@
+# Teaching
+
+Best practices and guides will be collected in tutorials and (virtual) coursework.
+This includes applications and support in providing routes to access and use compute resources for bio-structure predictions.
+
+In-person classes and seminars are published in our [news section](/blog).
+
+[Contact us!](mailto:ssc-biostructurehub@uni-heidelberg.de)
diff --git a/docs/stylesheets/extra.css b/docs/stylesheets/extra.css
@@ -9,9 +9,9 @@
 }
 
 .md-header__button.md-logo {
-    margin: 0;
+    margin: 1;
     padding: 1;
 }
 .md-header__button.md-logo img, .md-header__button.md-logo svg {
-    height: 2.3rem;
+    height: 3.5rem;
 }
diff --git a/docs/tutorials_alphafold.md → docs/tutorials/tutorial_AF_bwVisu.md b/docs/tutorials_alphafold.md → docs/tutorials/tutorial_AF_bwVisu.md
@@ -10,7 +10,7 @@ To start, get access to bwVisu via bwForCluster Helix or SDS. For more informati
 
 [https://www.urz.uni-heidelberg.de/en/service-catalogue/software-and-applications/bwvisu](https://www.urz.uni-heidelberg.de/en/service-catalogue/software-and-applications/bwvisu) 
 
-For technical questions regarding the high performance cluster, see [https://bw-support.scc.kit.edu](https://bw-support.scc.kit.edu). Feel free to [contact us](/contact) for support.
+For technical questions regarding the high performance cluster, see [https://bw-support.scc.kit.edu](https://bw-support.scc.kit.edu). Feel free to [contact us](../contact.md) for support.
 
 ### Step 2: Obtain Model Weights from AlphaFold 
 
@@ -20,7 +20,9 @@ Each user needs to individually obtain the model weights for AlphaFold3. Downloa
 
 Note that this can take up to a few days!
 
-**Please note that your use of AlphaFold is subject to the terms and conditions outlined in the [AlphaFold Terms of Use](https://github.com/google-deepmind/alphafold3/blob/main/WEIGHTS_TERMS_OF_USE.md). You are responsible for ensuring you comply with these terms.**
+!!! danger  "Legal Note"
+
+    Please note that your use of AlphaFold is subject to the terms and conditions outlined in the [AlphaFold Terms of Use](https://github.com/google-deepmind/alphafold3/blob/main/WEIGHTS_TERMS_OF_USE.md). You are responsible for ensuring you comply with these terms.
 
 ### Step 3: Connect to bwVisu and Start Jupyter 
 
@@ -34,20 +36,20 @@ The first step of the AlphaFold prediction is a multi-sequence alignment (MSA).
 
 For the MSA step, select 8 CPU cores with 10 GB of memory. The GPU necessary for the second step will be requested later. 
 
-![Screenshot](images/tutorial/bwVisu_CPU.png)
+![Screenshot](../images/tutorial/bwVisu_CPU.png)
 <!--{: style="height:500px;width:750px"}-->
 
 Click on "Launch". This will bring you to a new screen showing your interactive sessions. Wait for your session to be ready, then click on "Connect to Jupyter". This brings you into a JupyterLab environment.
 
 Upload the notebooks from our [github](https://github.com/ssciwr/BioStructureHub/tree/main/notebooks) by clicking on the upload button:
 
-![Screenshot](images/tutorial/bwVisu_upload.png){: style="height:111px;width:444px"}
+![Screenshot](../images/tutorial/bwVisu_upload.png){: style="height:111px;width:444px"}
 
 After the upload, you can see the notebooks in the file browser on the left.
 
 The alphafold parameters need to be uploaded as well. The parameter file is zipped as `af3.bin.zst`. Unpack the file to obtain `af3.bin`. This file then needs to be uploaded to a directory in your home, such as `/af3models`. 
 
-![Screenshot](images/tutorial/bwVisu_Afold_params.png){: style="height:95px;width:268px"}
+![Screenshot](../images/tutorial/bwVisu_Afold_params.png){: style="height:95px;width:268px"}
 
 
 
@@ -69,7 +71,7 @@ Decide where you want your working directory and output files to be:
 
 These directories can be created by clicking on the folder icon on the top left:
 
-![Screenshot](images/tutorial/bwVisu_newDir.png){: style="height:111px;width:444px"}
+![Screenshot](../images/tutorial/bwVisu_newDir.png){: style="height:111px;width:444px"}
 
 
 #### Prepare Input File 
@@ -84,7 +86,7 @@ Important parameters in the input file are the `name`, `sequence` and `id`, whic
 
 Next, we need to tell the AlphaFold3 program what to do with the input file, where to find the model weight parameters and where to write the output. Execute the next cell to write the run file that controls the execution. You don't need to worry about the parameters too much. They are prepared for you. Only change them if you know what you're doing. 
 
-![Screenshot](images/tutorial/bwVisu_Afold_MSA_input.png){: style="height:112px;width:268px"}
+![Screenshot](../images/tutorial/bwVisu_Afold_MSA_input.png){: style="height:112px;width:268px"}
 
 #### Run MSA Prediction 
 
@@ -95,13 +97,13 @@ Run the MSA prediction by executing the next cell:
 
 This will take about 5-10 minutes, but eventually, you should see... 
 
-![Screenshot](images/tutorial/bwVisu_Afold_MSA_done.png){: style="height:53px;width:379px"}
+![Screenshot](../images/tutorial/bwVisu_Afold_MSA_done.png){: style="height:53px;width:379px"}
 
 #### Verify Output 
 
 In the output directory, there should be a second `.json` file in the `output/test` directory. This includes all the information from the input file and the results of the MSA. 
 
-![Screenshot](images/tutorial/bwVisu_Afold_json.png)
+![Screenshot](../images/tutorial/bwVisu_Afold_json.png)
 {: style="height:89px;width:268px"}
 
 
@@ -115,16 +117,17 @@ The second step of the AlphaFold prediction is the inference of the structure by
 
 For the inference step we need a GPU, so we need to request a GPU node on bwVisu. A list of available GPUs and their specifications is available at [https://wiki.bwhpc.de/e/Helix/Hardware#Compute_Nodes](https://wiki.bwhpc.de/e/Helix/Hardware#Compute_Nodes), or in the table below.
 
-![Screenshot](images/tutorial/Helix_GPU.png)
+![Screenshot](../images/tutorial/Helix_GPU.png)
 <!--Cant I link this directly?-->
 
 The GPU is selected by "GPU Type". The memory of each GPU Type is specified in GPU Memory per GPU (GB). For this example we select one of the A40 GPUs.
 
-![Screenshot](images/tutorial/bwVisu_GPU.png)
+![Screenshot](../images/tutorial/bwVisu_GPU.png)
 <!--{: style="height:500px;width:750px"}-->
 
 Larger jobs (= longer sequences, more chains) require more memory. To access these, it is suggested to run the job directly on the Helix cluster. We will prepare a tutorial for this shortly - feel free to contact us!
 
+
 ### Step 7: Set Up Your Diffusion Run Within the Notebook
 - dependencies are missing
  Open `AFold_Diffusion_GPU.ipynb`.
@@ -150,7 +153,7 @@ Decide where you want your output files to be:
 
 Next, we need to tell the AlphaFold3 program what to do in the second part. Execute the next cell to write the run file that controls the execution. You don't need to worry about the parameters too much. They are prepared for you. Only change them if you know what you're doing. 
 
-![Screenshot](images/tutorial/bwVisu_Afold_GPU_input.png)
+![Screenshot](../images/tutorial/bwVisu_Afold_GPU_input.png)
 {: style="height:159px;width:268px"}
 
 
@@ -162,14 +165,14 @@ Execute the next cells to run the alignment job. Good luck!
 
 This may take a few minutes, but eventually, you should see... 
 
-![Screenshot](images/tutorial/bwVisu_Afold_GPU_done.png)
+![Screenshot](../images/tutorial/bwVisu_Afold_GPU_done.png)
 {: style="height:55px;width:357px"}
 
 #### Verify Output
 
 You should see the AlphaFold output files:
 
-![Screenshot](images/tutorial/bwVisu_Afold_GPU_output.png)
+![Screenshot](../images/tutorial/bwVisu_Afold_GPU_output.png)
 {: style="height:335px;width:268px"}
 
 By default AlphaFold creates 5 samples from one seed, and sorts them in individual directories. Their ranking scores are reported in a csv table.
@@ -181,14 +184,10 @@ The best model is presented in the output directory as well, with its structure
 
 Open the last notebook `Afold_Confidence_Levels.ipynb` to get a summary of the models confidence levels. This notebook reads the confidence descriptions and renders its central information.
 
-For this last notebook, you need to install a few dependencies into your environment. These dependencies are libraries that are used to analyze and visualize the output. The dependencies are installed in the Jupyter notebook in the first code cell:
-
-    %pip install biopython seaborn
+For this last notebook, you need to have access to a shared directory that includes libraries that are used to analyze and visualize the output. Define the `Kernel Path` to the AlphaFold kernel at `/mnt/sds-hd/sd25g005/afold3/share/jupyter/`. [Contact us](/contact.md) for access to this shared directory.
 
-After installing the dependencies, you need to restart the Jupyter kernel so that Jupyter finds the newly installed packages. Click on the circular arrow in the top left of the Jupyter notebook toolbar.
-
-![Screenshot](images/tutorial/restart_kernel.png)
-{: style="width:268px"}
+![Screenshot](../images/tutorial/bwVisu_GPU_Kernel.png)
+<!--{: style="height:500px;width:750px"}-->
 
 After this, the analysis should run without any errors. Explanations of the output are provided in the notebook.