OpenBioML · kjappelbaum · Nov 1, 2023 · Oct 18, 2023 · Oct 18, 2023 · Oct 18, 2023
diff --git a/data/tabular/SIDER/meta.yaml b/data/tabular/SIDER/meta.yaml
@@ -0,0 +1,223 @@
+---
+name: SIDER
+description: Database of marketed drugs and adverse drug reactions (ADR), grouped into 24 system organ classes.
+identifiers:
+    - id: SMILES
+      type: SMILES
+      description: SMILES
+targets:
+    - id: hepatobiliary_disorders
+      description: hepatobiliary disorders
+      type: boolean
+      names:
+          - noun: hepatobiliary disorders
+          - noun: liver and gallbladder disorders
+    - id: metabolism_and_nutrition_disorders
+      description: metabolism and nutrition disorders
+      type: boolean
+      names:
+          - noun: metabolism and nutrition disorders
+          - noun: metabolic and nutritional disorders
+    - id: eye_disorders
+      description: eye disorders
+      type: boolean
+      names:
+          - noun: eye disorders
+          - noun: ophthalmic disorders
+    - id: musculoskeletal_and_connective_tissue_disorders
+      description: musculoskeletal and connective tissue disorders
+      type: boolean
+      names:
+          - noun: musculoskeletal and connective tissue disorders
+          - noun: muscle and joint disorders
+    - id: gastrointestinal_disorders
+      description: gastrointestinal disorders
+      type: boolean
+      names:
+          - noun: gastrointestinal disorders
+          - noun: digestive system disorders
+    - id: immune_system_disorders
+      description: immune system disorders
+      type: boolean
+      names:
+          - noun: immune system disorders
+          - noun: disorders of the immune system
+    - id: reproductive_system_and_breast_disorders
+      description: reproductive system and breast disorders
+      type: boolean
+      names:
+          - noun: reproductive system and breast disorders
+          - noun: disorders of the breasts and the reproductive system
+    - id: neoplasms_benign_malignant_and_unspecified_(incl_cysts_and_polyps)
+      description: neoplasms benign, malignant and unspecified (incl cysts and polyps)
+      type: boolean
+      names:
+          - noun: neoplasms benign, malignant and unspecified (incl cysts and polyps)
+          - noun: benign and malignant tumors (including cysts and polyps)
+    - id: general_disorders_and_administration_site_conditions
+      description: general disorders and administration site conditions
+      type: boolean
+      names:
+          - noun: general disorders and administration site conditions
+          - noun: general health and administration site conditions
+    - id: endocrine_disorders
+      description: endocrine disorders
+      type: boolean
+      names:
+          - noun: endocrine disorders
+          - noun: endocrine system disorders
+    - id: surgical_and_medical_procedures
+      description: surgical and medical procedures
+      type: boolean
+      names:
+          - noun: surgical and medical procedures
+          - noun: medical and surgical procedures
+    - id: vascular_disorders
+      description: vascular disorders
+      type: boolean
+      names:
+          - noun: vascular disorders
+          - noun: vascular system disorders
+    - id: blood_and_lymphatic_system_disorders
+      description: blood and lymphatic system disorders
+      type: boolean
+      names:
+          - noun: blood and lymphatic system disorders
+          - noun: disorders of the blood and lymphatic system
+    - id: skin_and_subcutaneous_tissue_disorders
+      description: skin and subcutaneous tissue disorders
+      type: boolean
+      names:
+          - noun: skin and subcutaneous tissue disorders
+          - noun: disorders of the skin and subcutaneous tissue
+    - id: congenital_familial_and_genetic_disorders
+      description: congenital, familial and genetic disorders
+      type: boolean
+      names:
+          - noun: congenital, familial and genetic disorders
+          - noun: familial, congenital and genetic disorders
+    - id: infections_and_infestations
+      description: infections and infestations
+      type: boolean
+      names:
+          - noun: infections and infestations
+          - noun: infestations and infections
+    - id: respiratory_thoracic_and_mediastinal_disorders
+      description: respiratory, thoracic and mediastinal disorders
+      type: boolean
+      names:
+          - noun: respiratory, thoracic and mediastinal disorders
+          - noun: respiratory and thoracic disorders
+    - id: psychiatric_disorders
+      description: psychiatric disorders
+      type: boolean
+      names:
+          - noun: psychiatric disorders
+          - noun: mental health and psychiatric disorders
+    - id: renal_and_urinary_disorders
+      description: renal and urinary disorders
+      type: boolean
+      names:
+          - noun: renal and urinary disorders
+          - noun: kidney and urinary tract disorders
+    - id: pregnancy_puerperium_and_perinatal_conditions
+      description: pregnancy, puerperium and perinatal conditions
+      type: boolean
+      names:
+          - noun: pregnancy, puerperium and perinatal conditions
+          - noun: pregnancy, childbirth, and newborn conditions
+    - id: ear_and_labyrinth_disorders
+      description: ear and labyrinth disorders
+      type: boolean
+      names:
+          - noun: ear and labyrinth disorders
+          - noun: ear and inner ear disorders
+    - id: cardiac_disorders
+      description: cardiac disorders
+      type: boolean
+      names:
+          - noun: cardiac disorders
+          - noun: cardiovascular disorders
+    - id: nervous_system_disorders
+      description: nervous system disorders
+      type: boolean
+      names:
+          - noun: nervous system disorders
+          - noun: disorders of the nervous system
+    - id: injury_poisoning_and_procedural_complications
+      description: injury, poisoning and procedural complications
+      type: boolean
+      names:
+          - noun: injury, poisoning and procedural complications
+          - noun: injuries and poisonings
+license: CC BY 4.0
+links:
+    - url: https://academic.oup.com/nar/article/44/D1/D1075/2502602?login=false
+      description: corresponding publication
+    - url: https://deepchemdata.s3-us-west-1.amazonaws.com/datasets/sider.csv.gz
+      description: Data source
+num_points: 1427
+bibtex:
+    - |-
+      @article{10.1093/nar/gkv1075,
+      author = {Kuhn, Michael and Letunic, Ivica and Jensen, Lars Juhl and Bork, Peer},
+      title = "{The SIDER database of drugs and side effects}",
+      journal = {Nucleic Acids Research},
+      volume = {44},
+      number = {D1},
+      pages = {D1075-D1079},
+      year = {2015},
+      month = {10},
+      issn = {0305-1048},
+      doi = {10.1093/nar/gkv1075},
+      url = {https://doi.org/10.1093/nar/gkv1075},
+      }
+templates:
+    - The {#molecule|compound|chemical|molecular species|chemical compound!} with the {SMILES__description} {#representation of |!}{SMILES#} is {hepatobiliary_disorders#not
+      a &a }{#potential cause|potential reason!} for {hepatobiliary_disorders__names__noun}.
+    - The {#molecule|compound|chemical|molecular species|chemical compound!} with the {SMILES__description} {#representation of |!}{SMILES#} is {metabolism_and_nutrition_disorders#not
+      a &a }{#potential cause|potential reason!} for {metabolism_and_nutrition_disorders__names__noun}.
+    - The {#molecule|compound|chemical|molecular species|chemical compound!} with the {SMILES__description} {#representation of |!}{SMILES#} is {eye_disorders#not
+      a &a }{#potential cause|potential reason!} for {eye_disorders__names__noun}.
+    - The {#molecule|compound|chemical|molecular species|chemical compound!} with the {SMILES__description} {#representation of |!}{SMILES#} is {musculoskeletal_and_connective_tissue_disorders#not
+      a &a }{#potential cause|potential reason!} for {musculoskeletal_and_connective_tissue_disorders__names__noun}.
+    - The {#molecule|compound|chemical|molecular species|chemical compound!} with the {SMILES__description} {#representation of |!}{SMILES#} is {gastrointestinal_disorders#not
+      a &a }{#potential cause|potential reason!} for {gastrointestinal_disorders__names__noun}.
+    - The {#molecule|compound|chemical|molecular species|chemical compound!} with the {SMILES__description} {#representation of |!}{SMILES#} is {immune_system_disorders#not
+      a &a }{#potential cause|potential reason!} for {immune_system_disorders__names__noun}.
+    - The {#molecule|compound|chemical|molecular species|chemical compound!} with the {SMILES__description} {#representation of |!}{SMILES#} is {reproductive_system_and_breast_disorders#not
+      a &a }{#potential cause|potential reason!} for {reproductive_system_and_breast_disorders__names__noun}.
+    - The {#molecule|compound|chemical|molecular species|chemical compound!} with the {SMILES__description} {#representation of |!}{SMILES#} is {neoplasms_benign_malignant_and_unspecified_(incl_cysts_and_polyps)#not
+      a &a }{#potential cause|potential reason!} for {neoplasms_benign_malignant_and_unspecified_(incl_cysts_and_polyps)__names__noun}.
+    - The {#molecule|compound|chemical|molecular species|chemical compound!} with the {SMILES__description} {#representation of |!}{SMILES#} is {general_disorders_and_administration_site_conditions#not
+      a &a }{#potential cause|potential reason!} for {general_disorders_and_administration_site_conditions__names__noun}.
+    - The {#molecule|compound|chemical|molecular species|chemical compound!} with the {SMILES__description} {#representation of |!}{SMILES#} is {endocrine_disorders#not
+      a &a }{#potential cause|potential reason!} for {endocrine_disorders__names__noun}.
+    - The {#molecule|compound|chemical|molecular species|chemical compound!} with the {SMILES__description} {#representation of |!}{SMILES#} is {surgical_and_medical_procedures#not
+      a &a }{#potential cause|potential reason!} for {surgical_and_medical_procedures__names__noun}.
+    - The {#molecule|compound|chemical|molecular species|chemical compound!} with the {SMILES__description} {#representation of |!}{SMILES#} is {vascular_disorders#not
+      a &a }{#potential cause|potential reason!} for {vascular_disorders__names__noun}.
+    - The {#molecule|compound|chemical|molecular species|chemical compound!} with the {SMILES__description} {#representation of |!}{SMILES#} is {blood_and_lymphatic_system_disorders#not
+      a &a }{#potential cause|potential reason!} for {blood_and_lymphatic_system_disorders__names__noun}.
+    - The {#molecule|compound|chemical|molecular species|chemical compound!} with the {SMILES__description} {#representation of |!}{SMILES#} is {skin_and_subcutaneous_tissue_disorders#not
+      a &a }{#potential cause|potential reason!} for {skin_and_subcutaneous_tissue_disorders__names__noun}.
+    - The {#molecule|compound|chemical|molecular species|chemical compound!} with the {SMILES__description} {#representation of |!}{SMILES#} is {congenital_familial_and_genetic_disorders#not
+      a &a }{#potential cause|potential reason!} for {congenital_familial_and_genetic_disorders__names__noun}.
+    - The {#molecule|compound|chemical|molecular species|chemical compound!} with the {SMILES__description} {#representation of |!}{SMILES#} is {infections_and_infestations#not
+      a &a }{#potential cause|potential reason!} for {infections_and_infestations__names__noun}.
+    - The {#molecule|compound|chemical|molecular species|chemical compound!} with the {SMILES__description} {#representation of |!}{SMILES#} is {respiratory_thoracic_and_mediastinal_disorders#not
+      a &a }{#potential cause|potential reason!} for {respiratory_thoracic_and_mediastinal_disorders__names__noun}.
+    - The {#molecule|compound|chemical|molecular species|chemical compound!} with the {SMILES__description} {#representation of |!}{SMILES#} is {psychiatric_disorders#not
+      a &a }{#potential cause|potential reason!} for {psychiatric_disorders__names__noun}.
+    - The {#molecule|compound|chemical|molecular species|chemical compound!} with the {SMILES__description} {#representation of |!}{SMILES#} is {renal_and_urinary_disorders#not
+      a &a }{#potential cause|potential reason!} for {renal_and_urinary_disorders__names__noun}.
+    - The {#molecule|compound|chemical|molecular species|chemical compound!} with the {SMILES__description} {#representation of |!}{SMILES#} is {pregnancy_puerperium_and_perinatal_conditions#not
+      a &a }{#potential cause|potential reason!} for {pregnancy_puerperium_and_perinatal_conditions__names__noun}.
+    - The {#molecule|compound|chemical|molecular species|chemical compound!} with the {SMILES__description} {#representation of |!}{SMILES#} is {ear_and_labyrinth_disorders#not
+      a &a }{#potential cause|potential reason!} for {ear_and_labyrinth_disorders__names__noun}.
+    - The {#molecule|compound|chemical|molecular species|chemical compound!} with the {SMILES__description} {#representation of |!}{SMILES#} is {cardiac_disorders#not
+      a &a }{#potential cause|potential reason!} for {cardiac_disorders__names__noun}.
+    - The {#molecule|compound|chemical|molecular species|chemical compound!} with the {SMILES__description} {#representation of |!}{SMILES#} is {nervous_system_disorders#not
+      a &a }{#potential cause|potential reason!} for {nervous_system_disorders__names__noun}.
+    - The {#molecule|compound|chemical|molecular species|chemical compound!} with the {SMILES__description} {#representation of |!}{SMILES#} is {injury_poisoning_and_procedural_complications#not
+      a &a }{#potential cause|potential reason!} for {injury_poisoning_and_procedural_complications__names__noun}.
diff --git a/data/tabular/SIDER/transform.py b/data/tabular/SIDER/transform.py
@@ -0,0 +1,148 @@
+from typing import List, Tuple
+
+import pandas as pd
+import yaml
+
+ALT_DESCRIPTIONS = [
+    "Liver and Gallbladder Disorders",
+    "Metabolic and Nutritional Disorders",
+    "Ophthalmic Disorders",
+    "Muscle and Joint Disorders",
+    "Digestive System Disorders",
+    "Disorders of the Immune System",
+    "Disorders of the breasts and the Reproductive system",
+    "Benign and Malignant Tumors (including Cysts and Polyps)",
+    "General Health and Administration Site Conditions",
+    "Endocrine System Disorders",
+    "Medical and Surgical Procedures",
+    "Vascular System Disorders",
+    "Disorders of the blood and lymphatic system",
+    "Disorders of the Skin and Subcutaneous Tissue",
+    "Familial, Congenital and Genetic Disorders",
+    "Infestations and Infections",
+    "Respiratory and Thoracic Disorders",
+    "Mental Health and Psychiatric Disorders",
+    "Kidney and Urinary Tract Disorders",
+    "Pregnancy, Childbirth, and Newborn Conditions",
+    "Ear and Inner Ear Disorders",
+    "Cardiovascular Disorders",
+    "Disorders of the Nervous System",
+    "Injuries and Poisonings",
+]
+
+
+def load_dataset() -> pd.DataFrame:
+    sider = pd.read_csv(
+        "https://deepchemdata.s3-us-west-1.amazonaws.com/datasets/sider.csv.gz"
+    )
+    sider = sider.drop(
+        columns=["Product issues", "Social circumstances", "Investigations"]
+    )
+    return sider
+
+
+def transform_data() -> Tuple[pd.DataFrame, pd.Index]:
+    sider = load_dataset()
+    old_columns = sider.columns.str.lower()
+    sider.columns = sider.columns.str.lower().str.replace(" ", "_").str.replace(",", "")
+    sider = sider.rename(columns={"smiles": "SMILES"})
+    sider.to_csv("data_clean.csv", index=False)
+
+    return sider, old_columns
+
+
+def write_meta(column_ids: pd.Index, descriptions: List[str], num_points: int) -> None:
+    # Write metadata
+    targets = [
+        {
+            "id": f"{col_id}",
+            "description": f"{description}",
+            "type": "boolean",
+            "names": [
+                {"noun": f"{description}".lower()},
+                {"noun": f"{alt_desc}".lower()},
+            ],
+        }
+        for col_id, description, alt_desc in zip(
+            column_ids[1:], descriptions[1:], ALT_DESCRIPTIONS
+        )
+    ]
+
+    templates = [
+        "The {#molecule|compound|chemical|molecular species|chemical compound!} with the {SMILES__description}"  # noqa: E501
+        + " {#representation of |!}{SMILES#} is {"
+        + col_id
+        + "#not a &a }"
+        + "{#potential cause|potential reason!} for {"
+        + col_id
+        + "__names__noun}."  # noqa: E501
+        for col_id in column_ids[1:]
+    ]
+
+    meta = {
+        "name": "SIDER",  # unique identifier, we will also use this for directory names
+        "description": f"""Database of marketed drugs and adverse drug reactions (ADR), grouped into {len(column_ids[1:])} system organ classes.""",  # noqa: E501
+        "identifiers": [
+            {
+                "id": "SMILES",  # column name
+                "type": "SMILES",
+                "description": "SMILES",  # description (optional, except for "Other")
+            }
+        ],
+        "targets": targets,
+        "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://academic.oup.com/nar/article/44/D1/D1075/2502602?login=false",
+                "description": "corresponding publication",
+            },
+            {
+                "url": "https://deepchemdata.s3-us-west-1.amazonaws.com/datasets/sider.csv.gz",
+                "description": "Data source",
+            },
+        ],
+        "num_points": num_points,  # number of datapoints in this dataset
+        "bibtex": [
+            """@article{10.1093/nar/gkv1075,
+author = {Kuhn, Michael and Letunic, Ivica and Jensen, Lars Juhl and Bork, Peer},
+title = "{The SIDER database of drugs and side effects}",
+journal = {Nucleic Acids Research},
+volume = {44},
+number = {D1},
+pages = {D1075-D1079},
+year = {2015},
+month = {10},
+issn = {0305-1048},
+doi = {10.1093/nar/gkv1075},
+url = {https://doi.org/10.1093/nar/gkv1075},
+}""",
+        ],
+        "templates": templates,
+    }
+
+    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!")
+
+
+def main():
+    sider, old_columns = transform_data()
+    write_meta(sider.columns, old_columns, len(sider))
+
+
+if __name__ == "__main__":
+    main()