Small changes to about tab

app.py

@@ -66,7 +66,7 @@ def update_current_dataframe():
         logger.info("Fetching latest dataset for leaderboard...")
         current_df = fetch_dataset_df()
         logger.debug(f"Dataset version updated")
-        time.sleep(30)  # Check for updates every 30 seconds
+        time.sleep(60)  # Check for updates every 60 seconds
 
 threading.Thread(target=update_current_dataframe, daemon=True).start()
 
@@ -138,11 +138,13 @@ with gr.Blocks(title="OpenADMET ADMET Challenge", fill_height=False,
     - Mouse Gastrocnemius Muscle Binding (**MGMB**): % Unbound
     
     Find more information about these endpoints on our [blog](https://openadmet.ghost.io/openadmet-expansionrx-blind-challenge/).
-    **UPDATE:** The Challenge is now live! Data available at the following Hugging Face Datasets 
-    Training: https://huggingface.co/datasets/openadmet/openadmet-expansionrx-challenge-train-data
-    Test: https://huggingface.co/datasets/openadmet/openadmet-expansionrx-challenge-test-data-blinded
-    You can also watch a [Webinar](https://www.youtube.com/watch?v=9v0Ej_FL6k0) introducing the challenge run with [Collaborative Drug Discovery](https://www.collaborativedrug.com/).
-    We also have a [form](https://forms.gle/KiviZ7AaGcuqtrwH8) you can fill out for access to a CDD vault containing the challenge data and access to some other tools.
+    
+    **UPDATE:** The Challenge is now live! Data available at the following Hugging Face Datasets
+
+    - Training: https://huggingface.co/datasets/openadmet/openadmet-expansionrx-challenge-train-data
+    - Test: https://huggingface.co/datasets/openadmet/openadmet-expansionrx-challenge-test-data-blinded
+    
+    You can also watch a [Webinar](https://www.youtube.com/watch?v=9v0Ej_FL6k0) where we introduce the challenge, hosted by [Collaborative Drug Discovery (CDD)](https://www.collaborativedrug.com/).
     
     ## ✅ How to Participate
     1. **Register**: Create an account with Hugging Face.
@@ -166,12 +168,15 @@ with gr.Blocks(title="OpenADMET ADMET Challenge", fill_height=False,
     | Caco-2 Permeability Papp A>B | 10^-6 cm/s  |   float   | Caco-2 Permeability Papp A>B |
     | MPPB                         | % Unbound   |   float   | Mouse Plasma Protein Binding |
     | MBPB                         | % Unbound   |   float   | Mouse Brain Protein Binding |
-    | MGMB.                        | % Unbound   |   float   | Mouse Gastrocnemius Muscle Binding |
+    | MGMB                         | % Unbound   |   float   | Mouse Gastrocnemius Muscle Binding |
+ 
     You can download the training data from the [Hugging Face dataset](https://huggingface.co/datasets/openadmet/openadmet-challenge-train-data).
+
     The test set will remained blinded until the challenge submission deadline. You will be tasked with predicting the same set of ADMET endpoints for the test set molecules.
     
-    The training and blinded test set will also be made available on the [CDD Vault](https://www.collaborativedrug.com/). An account to access the CDD Vault can be requested by emailing **openadmet@omsf.io**. 
+    The training and blinded test set will also be made available on the [CDD Vault](https://www.collaborativedrug.com/). An account to access the CDD Vault can be requested by filling out this [form](https://forms.gle/KiviZ7AaGcuqtrwH8, which can also be used to request access to some other tools.
     Note that by joining the Vault, your account will be visible to other participants, so this option is **not recommended for those wishing to remain anonymous.**
+
     ## 📝 Evaluation
     The challenge will be judged based on the following criteria:
     - We welcome submissions of any kind, including machine learning and physics-based approaches. You can also employ pre-training approaches as you see fit, 
@@ -183,7 +188,8 @@ with gr.Blocks(title="OpenADMET ADMET Challenge", fill_height=False,
     - The endpoints will be judged individually by mean absolute error (**MAE**), while an overall leaderboard will be judged by the macro-averaged relative absolute error (**MA-RAE**). 
     - For endpoints that are not already on a log scale (e.g LogD) they will be transformed to log scale to minimize the impact of outliers on evaluation.
     - We will estimate errors on the metrics using bootstrapping and use the statistical testing workflow outlined in [this paper](https://chemrxiv.org/engage/chemrxiv/article-details/672a91bd7be152b1d01a926b) to determine if model performance is statistically distinct.
-    📅 **Timeline**:  
+    
+    ## 📅 **Timeline**:  
     - **September 16:** Challenge announcement
     - **October 14:** Second announcement and sample data release
     - **October 27:** Challenge starts

requirements.txt

@@ -5,4 +5,5 @@ gradio-leaderboard
 plotly
 scipy
 scikit-learn
-loguru
+loguru
+statsmodels

utils.py

@@ -11,6 +11,8 @@ def make_user_clickable(name: str):
     link =f'https://huggingface.co/{name}'
     return f'<a target="_blank" href="{link}" style="color: var(--link-text-color); text-decoration: underline;text-decoration-style: dotted;">{name}</a>'
 def make_tag_clickable(tag: str):
+    if tag is None:
+        return "Not submitted"
     return f'<a target="_blank" href="{tag}" style="color: var(--link-text-color); text-decoration: underline;text-decoration-style: dotted;">link</a>'
 
 def fetch_dataset_df():