Upload app.py

app.py

@@ -868,9 +868,9 @@ To predict interactions/binding affinities of a single target against a library
                             drug_library = gr.Dropdown(label='Step 3. Select or Upload a Compound Library',
                                                        choices=list(DRUG_LIBRARY_MAP.keys()))
                             with gr.Row():
-                                gr.File(label='Example SDF compound library',
+                                gr.File(label='Example SDF Compound Library',
                                         value='data/examples/compound_library.sdf', interactive=False)
-                                gr.File(label='Example CSV compound library',
+                                gr.File(label='Example CSV Compound Library',
                                         value='data/examples/compound_library.csv', interactive=False)
                             drug_library_upload_btn = gr.UploadButton(
                                 label='Upload a custom library', variant='primary')
@@ -918,19 +918,6 @@ To predict interactions/binding affinities of a single target against a library
 <center>
 To predict interactions/binding affinities of a single compound against a library of protein targets.
 </center>
-
-ℹ️ A custom target library can be a FASTA file with a single or multiple amino acid sequences, 
-or a CSV file has a required FASTA string column and optionally an ID column:
-
-<b>X2</b>: the FASTA sequence of a target\n
-<b>ID2</b> (optional): the ID (PubChem or any arbitrary unique identifier) of a compound\n
-
-Example CSV target library:
-
-| X2            | ID2    |
-|---------------|--------|
-| MVQKSRNGGV... | O88943 |
-| MTSPSSSPVF... | Q9Y5S1 |
                 ''')
             with gr.Blocks() as identify_block:
                 with gr.Column() as identify_page:
@@ -971,9 +958,9 @@ Example CSV target library:
                             target_library = gr.Dropdown(label='Step 3. Select or Upload a Target Library',
                                                          choices=list(TARGET_LIBRARY_MAP.keys()))
                             with gr.Row():
-                                gr.File(label='Example FASTA target library',
+                                gr.File(label='Example FASTA Target Library',
                                         value='data/examples/target_library.fasta', interactive=False)
-                                gr.File(label='Example CSV target library',
+                                gr.File(label='Example CSV Target Library',
                                         value='data/examples/target_library.csv', interactive=False)
                             target_library_upload_btn = gr.UploadButton(
                                 label='Upload a custom library', variant='primary')
@@ -1016,64 +1003,72 @@ Example CSV target library:
             gr.Markdown('''
 # <center>DeepSEQreen Interaction Pair Inference</center>
 <center>To predict interactions/binding affinities between any compound-protein pairs.</center>
-
-ℹ️ A custom interaction pair dataset can be generated from a FASTA file containing multiple sequences 
-and a SDF file containing multiple compounds (for predicting CPI/CPA of all possible combinations of 
-compound-protein pairs), or a CSV file with 2 required string columns and optionally 2 ID columns:
-
-<b>X1</b>: the SMILES string of a compound\n
-<b>X2</b>: the FASTA sequence of a target\n
-<b>ID1</b> (optional): the ID (PubChem or any arbitrary unique identifier) of a compound\n
-<b>ID2</b> (optional): the ID (UniProt or any arbitrary unique identifier) of a protein
-
-Example CSV interaction pair dataset:
-
-| X1                                      | X2            | ID1          | ID2    |
-|---------------------------------------- |---------------|--------------|--------|
-| CCOC(=O)Nc1ccc(NCc2ccc(F)cc2)cc1N       | MVQKSRNGGV... | CHEMBL41355  | O88943 |
-| CCCCCc1cc(O)c(C/C=C(\C)CCC=C(C)C)c(O)c1 | MTSPSSSPVF... | CHEMBL497318 | Q9Y5S1 |
 ''')
             with gr.Blocks() as infer_block:
                 with gr.Column() as infer_page:
-                    infer_type = gr.Dropdown(choices=['Upload a compound library and a target library',
-                                                      'Upload a CSV interaction pair dataset'],
-                                             value='Upload a compound library and a target library')
+                    infer_type = gr.Dropdown(choices=['Upload a CSV interaction pair dataset',
+                                                      'Upload a compound library and a target library'],
+                                             label='Step 1. Select Pair Input Type and Input',
+                                             value='Upload a CSV interaction pair dataset')
                     with gr.Column() as pair_upload:
-                        gr.File(label="Example custom dataset",
-                                value="data/examples/interaction_pair_inference.csv",
-                                interactive=False)
-                        with gr.Column():
+                        with gr.Row():
+                            gr.File(label="Example custom dataset",
+                                    value="data/examples/interaction_pair_inference.csv",
+                                    interactive=False)
+                        with gr.Row():
                             infer_data_for_predict = gr.File(
-                                label='Upload a custom dataset', file_count="single", type='filepath', visible=True)
+                                label='Upload a Custom Dataset', file_count="single", type='filepath', visible=True)
                     with gr.Column() as pair_generate:
                         with gr.Row():
-                            gr.File(label='Example SDF compound library',
+                            gr.File(label='Example SDF Compound Library',
                                     value='data/examples/compound_library.sdf', interactive=False)
-                            gr.File(label='Example FASTA target library',
+                            gr.File(label='Example FASTA Target Library',
                                     value='data/examples/target_library.fasta', interactive=False)
                         with gr.Row():
-                            gr.File(label='Example CSV compound library',
+                            gr.File(label='Example CSV Compound Library',
                                     value='data/examples/compound_library.csv', interactive=False)
-                            gr.File(label='Example CSV target library',
+                            gr.File(label='Example CSV Target Library',
                                     value='data/examples/target_library.csv', interactive=False)
                         with gr.Row():
-                            infer_drug = gr.File(label='SDF/CSV file containing multiple compounds',
+                            infer_drug = gr.File(label='SDF/CSV File Containing Multiple Compounds',
                                                  file_count="single", type='filepath')
-                            infer_target = gr.File(label='FASTA/CSV file containing multiple targets',
+                            infer_target = gr.File(label='FASTA/CSV File Containing Multiple Targets',
                                                    file_count="single", type='filepath')
 
-                    with gr.Row(visible=True):
-                        pair_infer_task = gr.Dropdown(list(TASK_MAP.keys()), label='Task')
-                        pair_infer_preset = gr.Dropdown(list(PRESET_MAP.keys()), label='Preset')
-                        pair_infer_target_family = gr.Dropdown(choices=['General'],
-                                                               label='Target family',
-                                                               value='General')
+                    with gr.Row():
+                        with gr.Column():
+                            HelpTip(
+                                "By default, models trained on all protein families (general) will be applied."
+                                "If the proteins in the target library of interest all belong to the same protein family, manually selecting the family is supported."
+                            )
+                            pair_infer_target_family = gr.Dropdown(choices=list(TARGET_FAMILY_MAP.keys()),
+                                                                   value='General',
+                                                                   label='Step 2. Select Target Protein Family (Optional)')
 
-                    # with gr.Row():
-                    #     pair_infer_email = gr.Textbox(
-                    #         label='Email (optional)',
-                    #         info="Your email will be used to send you notifications when your job finishes."
-                    #     )
+                    with gr.Row():
+                        with gr.Column():
+                            HelpTip(
+                                "Interaction prediction provides you binding probability score between the target of interest and each compound in the library,"
+                                "while affinity prediction directly estimates their binding strength measured using IC50."
+                            )
+                            pair_infer_task = gr.Dropdown(list(TASK_MAP.keys()),
+                                                          label='Step 3. Select a Prediction Task',
+                                                          value='Compound-protein interaction')
+
+                    with gr.Row():
+                        with gr.Column():
+                            HelpTip("Select your preferred model, or click Recommend for the best-performing model based on the selected task, family, and random splitting validation."
+                                    "Please refer to documentation for detailed benchamrk results."
+                                    )
+                            pair_infer_preset = gr.Dropdown(list(PRESET_MAP.keys()), label='Step 4. Select a Preset Model')
+                            infer_preset_recommend_btn = gr.Button(value='Recommend a model', variant='primary')
+
+
+                    with gr.Row():
+                        pair_infer_email = gr.Textbox(
+                            label='Step 5. Email (Optional)',
+                            info="If an email is provided, a notification email will be sent to you when your job is completed."
+                        )
 
                     with gr.Row(visible=True):
                         # pair_infer_clr_btn = gr.ClearButton(size='lg')