add antibody developability from TDC

data/SAbDab_Chen/meta.yaml

@@ -0,0 +1,50 @@
+---
+name: SAbDab_Chen
+description: "Antibody data from Chen et al, where they process from the SAbDab. \n        From an initial dataset of 3816 antibodies, they retained 2426\
+    \ antibodies\n        that satisfy the following criteria: 1. \n        have both sequence (FASTA) and Protein Data Bank (PDB) structure files,\n  \
+    \      2. contain both a heavy chain and a light chain, and 3. \n        have crystal structures with resolution < 3 Å. \n        The DI label is derived\
+    \ from BIOVIA's pipelines."
+targets:
+    - id: developability
+      description: functional antibody candidate to be developed into a manufacturable(1), or not(0)
+      units: ''
+      type: categorical
+      names:
+          - antibody developability
+          - monoclonal anitbody
+          - functional antibody candidate
+          - manufacturable, stable, safe, and effective antibody drug
+      uris:
+          - https://rb.gy/idkdqp
+          - https://rb.gy/b8cx8i
+identifiers:
+    - id: antibody_pdb_ID
+      type: Other
+      description: anitbody pdb id
+    - id: heavy_chain
+      type: Other
+      description: anitbody heavy chain amino acid sequence in FASTA
+    - id: light_chain
+      type: Other
+      description: anitbody light chain amino acid sequence in FASTA
+license: CC BY 4.0
+links:
+    - url: https://doi.org/10.1101/2020.06.18.159798
+      description: corresponding publication
+    - url: https://doi.org/10.1093/nar/gkt1043
+      description: corresponding publication
+    - url: https://www.3ds.com/products-services/biovia/products/data-science/pipeline-pilot/
+      description: corresponding tools used
+    - url: https://tdcommons.ai/single_pred_tasks/develop/#sabdab-chen-et-al
+      description: data source
+num_points: 2409
+bibtex:
+    - "@article{Chen2020,\n          doi = {10.1101/2020.06.18.159798},\n          url = {https://doi.org/10.1101/2020.06.18.159798},\n          year =\
+      \ {2020},\n          month = jun,\n          publisher = {Cold Spring Harbor Laboratory},\n          author = {Xingyao Chen and Thomas Dougherty and\
+      \ \n          Chan Hong and Rachel Schibler and Yi Cong Zhao and \n          Reza Sadeghi and Naim Matasci and Yi-Chieh Wu and Ian Kerman},\n    \
+      \      title = {Predicting Antibody Developability from Sequence \n          using Machine Learning}}"
+    - "@article{Dunbar2013,\n          doi = {10.1093/nar/gkt1043},\n          url = {https://doi.org/10.1093/nar/gkt1043},\n          year = {2013},\n\
+      \          month = nov,\n          publisher = {Oxford University Press ({OUP})},\n          volume = {42},\n          number = {D1},\n          pages\
+      \ = {D1140--D1146},\n          author = {James Dunbar and Konrad Krawczyk and Jinwoo Leem \n          and Terry Baker and Angelika Fuchs and Guy Georges\
+      \ and Jiye Shi and\n          Charlotte M. Deane},\n          title = {{SAbDab}: the structural antibody database},\n          journal = {Nucleic\
+      \ Acids Research}}"

data/SAbDab_Chen/transform.py

@@ -0,0 +1,153 @@
+import pandas as pd
+import yaml
+from tdc.single_pred import Develop
+
+
+def get_and_transform_data():
+    # get raw data
+    target_folder = "SAbDab_Chen"
+    target_subfolder = "SAbDab_Chen"
+    data = Develop(name=target_subfolder)
+
+    # proceed raw data
+    df = data.get_data()
+    fields_orig = df.columns.tolist()
+    assert fields_orig == ["Antibody_ID", "Antibody", "Y"]
+
+    fn_data_original = "data_original.csv"
+
+    antibody_list = df.Antibody.tolist()
+    s2l = lambda list_string: list(
+        map(str.strip, list_string.strip("][").replace("'", "").split(","))
+    )
+    df["heavy_chain"] = [s2l(x)[0] for x in antibody_list]
+    df["light_chain"] = [s2l(x)[1] for x in antibody_list]
+    df = df[["Antibody_ID", "heavy_chain", "light_chain", "Y"]]
+    df.to_csv(fn_data_original, index=False)
+
+    #  load raw data and assert columns
+    df = pd.read_csv(fn_data_original, sep=",")
+    fields_orig = df.columns.tolist()
+    assert fields_orig == ["Antibody_ID", "heavy_chain", "light_chain", "Y"]
+    fields_clean = ["antibody_pdb_ID", "heavy_chain", "light_chain", "developability"]
+    df.columns = fields_clean
+    assert not df.duplicated().sum()
+
+    # save to csv
+    fn_data_csv = "data_clean.csv"
+    df.to_csv(fn_data_csv, index=False)
+
+    meta = {
+        "name": f"{target_folder}",  # unique identifier, we will also use this for directory names
+        "description": """Antibody data from Chen et al, where they process from the SAbDab.
+        From an initial dataset of 3816 antibodies, they retained 2426 antibodies
+        that satisfy the following criteria: 1.
+        have both sequence (FASTA) and Protein Data Bank (PDB) structure files,
+        2. contain both a heavy chain and a light chain, and 3.
+        have crystal structures with resolution < 3 Å.
+        The DI label is derived from BIOVIA's pipelines.""",
+        "targets": [
+            {
+                "id": "developability",  # name of the column in a tabular dataset
+                "description": "functional antibody candidate to be developed into a manufacturable(1), or not(0)",
+                "units": "",  # units of the values in this column (leave empty if unitless)
+                "type": "categorical",  # can be "categorical", "ordinal", "continuous"
+                "names": [  # names for the property (to sample from for building the prompts)
+                    "antibody developability",
+                    "monoclonal anitbody",
+                    "functional antibody candidate",
+                    "manufacturable, stable, safe, and effective antibody drug",
+                ],
+                "uris": [
+                    "https://rb.gy/idkdqp",
+                    "https://rb.gy/b8cx8i",
+                ],
+            },
+        ],
+        "identifiers": [
+            {
+                "id": "antibody_pdb_ID",  # column name
+                "type": "Other",  # can be "SMILES", "SELFIES", "IUPAC", "Other"
+                "description": "anitbody pdb id",  # description (optional, except for "Other")
+            },
+            {
+                "id": "heavy_chain",  # column name
+                "type": "Other",  # can be "SMILES", "SELFIES", "IUPAC", "Other"
+                "description": "anitbody heavy chain amino acid sequence in FASTA",  # description (optional, except for "Other")
+            },
+            {
+                "id": "light_chain",  # column name
+                "type": "Other",  # can be "SMILES", "SELFIES", "IUPAC", "Other"
+                "description": "anitbody light chain amino acid sequence in FASTA",  # description (optional, except for "Other")
+            },
+        ],
+        "license": "CC BY 4.0",  # license under which the original dataset was published
+        "links": [  # list of relevant links (original dataset, other uses, etc.)
+            {
+                "url": "https://doi.org/10.1101/2020.06.18.159798",
+                "description": "corresponding publication",
+            },
+            {
+                "url": "https://doi.org/10.1093/nar/gkt1043",
+                "description": "corresponding publication",
+            },
+            {
+                "url": "https://www.3ds.com/products-services/biovia/products/data-science/pipeline-pilot/",
+                "description": "corresponding tools used",
+            },
+            {
+                "url": "https://tdcommons.ai/single_pred_tasks/develop/#sabdab-chen-et-al",
+                "description": "data source",
+            },
+        ],
+        "num_points": len(df),  # number of datapoints in this dataset
+        "bibtex": [
+            """@article{Chen2020,
+          doi = {10.1101/2020.06.18.159798},
+          url = {https://doi.org/10.1101/2020.06.18.159798},
+          year = {2020},
+          month = jun,
+          publisher = {Cold Spring Harbor Laboratory},
+          author = {Xingyao Chen and Thomas Dougherty and
+          Chan Hong and Rachel Schibler and Yi Cong Zhao and
+          Reza Sadeghi and Naim Matasci and Yi-Chieh Wu and Ian Kerman},
+          title = {Predicting Antibody Developability from Sequence
+          using Machine Learning}}""",
+            """@article{Dunbar2013,
+          doi = {10.1093/nar/gkt1043},
+          url = {https://doi.org/10.1093/nar/gkt1043},
+          year = {2013},
+          month = nov,
+          publisher = {Oxford University Press ({OUP})},
+          volume = {42},
+          number = {D1},
+          pages = {D1140--D1146},
+          author = {James Dunbar and Konrad Krawczyk and Jinwoo Leem
+          and Terry Baker and Angelika Fuchs and Guy Georges and Jiye Shi and
+          Charlotte M. Deane},
+          title = {{SAbDab}: the structural antibody database},
+          journal = {Nucleic Acids Research}}""",
+        ],
+    }
+
+    def str_presenter(dumper, data):
+        """configures yaml for dumping multiline strings
+        Ref: https://stackoverflow.com/questions/8640959/how-can-i-control-what-scalar-form-pyyaml-uses-for-my-data
+        """
+        if data.count("\n") > 0:  # check for multiline string
+            return dumper.represent_scalar("tag:yaml.org,2002:str", data, style="|")
+        return dumper.represent_scalar("tag:yaml.org,2002:str", data)
+
+    yaml.add_representer(str, str_presenter)
+    yaml.representer.SafeRepresenter.add_representer(
+        str, str_presenter
+    )  # to use with safe_dum
+    fn_meta = "meta.yaml"
+    with open(fn_meta, "w") as f:
+        yaml.dump(meta, f, sort_keys=False)
+
+    print(f"Finished processing {meta['name']} dataset!")
+
+
+if __name__ == "__main__":
+    get_and_transform_data()

data/TAP/meta.yaml

@@ -0,0 +1,98 @@
+---
+name: TAP
+description: "Immunogenicity, instability, self-association, \n        high viscosity, polyspecificity, or poor expression can all preclude\n        an\
+    \ antibody from becoming a therapeutic. Early identification of these\n        negative characteristics is essential. Akin to the Lipinski guidelines,\n\
+    \        which measure druglikeness in small molecules, \n        Therapeutic Antibody Profiler (TAP) highlights antibodies \n        that possess characteristics\
+    \ that are rare/unseen in \n        clinical-stage mAb therapeutics."
+targets:
+    - id: CDR_Length
+      description: CDR Complementarity-determining regions length
+      units: ''
+      type: continuous
+      names:
+          - Antibody Complementarity-determining regions length
+          - Therapeutic Antibody Profiler
+          - antibody developability
+          - monoclonal anitbody
+      uris:
+          - https://rb.gy/s9gv88
+          - https://rb.gy/km77hq
+          - https://rb.gy/b8cx8i
+    - id: PSH
+      description: patches of surface hydrophobicity
+      units: ''
+      type: continuous
+      names:
+          - antibody patches of surface hydrophobicity
+          - Therapeutic Antibody Profiler
+          - antibody developability
+          - monoclonal anitbody
+      uris:
+          - https://rb.gy/bchhaa
+          - https://rb.gy/2irr4l
+          - https://rb.gy/b8cx8i
+    - id: PPC
+      description: patches of positive charge
+      units: ''
+      type: continuous
+      names:
+          - patches of positive charge
+          - Therapeutic Antibody Profiler
+          - antibody developability
+          - monoclonal anitbody
+      uris:
+          - https://rb.gy/b8cx8i
+    - id: PNC
+      description: patches of negative charge
+      units: ''
+      type: continuous
+      names:
+          - anitbody patches of negative charge
+          - Therapeutic Antibody Profiler
+          - antibody developability
+          - monoclonal anitbody
+      uris:
+          - https://rb.gy/b8cx8i
+    - id: SFvCSP
+      description: structural Fv charge symmetry parameter
+      units: ''
+      type: continuous
+      names:
+          - antibody structural Fv charge symmetry parameter
+          - Therapeutic Antibody Profiler
+          - antibody developability
+          - monoclonal anitbody
+      uris:
+          - https://rb.gy/uxyhc3
+          - https://rb.gy/b8cx8i
+identifiers:
+    - id: antibody_name
+      type: Other
+      description: anitbody name
+    - id: heavy_chain
+      type: Other
+      description: anitbody heavy chain amino acid sequence
+    - id: light_chain
+      type: Other
+      description: anitbody light chain amino acid sequence
+license: CC BY 4.0
+links:
+    - url: https://doi.org/10.1073/pnas.1810576116
+      description: corresponding publication
+    - url: https://tdcommons.ai/single_pred_tasks/develop/#tap
+      description: data source
+num_points: 241
+bibtex:
+    - |-
+      @article{Raybould2019,
+                doi = {10.1073/pnas.1810576116},
+                url = {https://doi.org/10.1073/pnas.1810576116},
+                year = {2019},
+                month = feb,
+                publisher = {Proceedings of the National Academy of Sciences},
+                volume = {116},
+                number = {10},
+                pages = {4025--4030},
+                author = {Matthew I. J. Raybould and Claire Marks and Konrad Krawczyk and Bruck Taddese and Jaroslaw Nowak and Alan P. Lewis and Alexander Bujotzek and Jiye Shi and Charlotte M. Deane},
+                title = {Five computational developability guidelines for therapeutic antibody profiling},
+                journal = {Proceedings of the National Academy of Sciences}}