Enable custom curation files

src/biomappings/cli.py

@@ -3,6 +3,7 @@
 """The biomappings CLI."""
 
 import sys
+from pathlib import Path
 
 import click
 from more_click import run_app
@@ -23,12 +24,25 @@ def main():
 if get_git_hash() is not None:
 
     @main.command()
-    @click.option("--path", type=click.Path(), help="A predictions TSV file path")
-    def web(path):
+    @click.option("--predictions-path", type=click.Path(), help="A predictions TSV file path")
+    @click.option("--positives-path", type=click.Path(), help="A positives curation TSV file path")
+    @click.option("--negatives-path", type=click.Path(), help="A negatives curation TSV file path")
+    @click.option("--unsure-path", type=click.Path(), help="An unsure curation TSV file path")
+    def web(
+        predictions_path: Path,
+        positives_path: Path,
+        negatives_path: Path,
+        unsure_path: Path,
+    ):
         """Run the biomappings web app."""
         from .wsgi import get_app
 
-        app = get_app(predictions_path=path)
+        app = get_app(
+            predictions_path=predictions_path,
+            positives_path=positives_path,
+            negatives_path=negatives_path,
+            unsure_path=unsure_path,
+        )
         run_app(app, with_gunicorn=False)
 
     @main.command()

src/biomappings/resources/__init__.py

@@ -4,7 +4,6 @@
 
 import csv
 import itertools as itt
-import os
 from collections import defaultdict
 from pathlib import Path
 from typing import (
@@ -167,9 +166,9 @@ def target_curie(self) -> str:
         return f"{self.target_prefix}:{self.target_identifier}"
 
 
-def get_resource_file_path(fname) -> str:
+def get_resource_file_path(fname) -> Path:
     """Get a resource by its file name."""
-    return os.path.join(RESOURCE_PATH, fname)
+    return RESOURCE_PATH.joinpath(fname)
 
 
 def _load_table(fname) -> List[Dict[str, str]]:
@@ -211,87 +210,108 @@ def mapping_sort_key(prediction: Mapping[str, str]) -> Tuple[str, ...]:
 TRUE_MAPPINGS_PATH = get_resource_file_path("mappings.tsv")
 
 
-def load_mappings() -> List[Dict[str, str]]:
+def load_mappings(*, path: Optional[Path] = None) -> List[Dict[str, str]]:
     """Load the mappings table."""
-    return _load_table(TRUE_MAPPINGS_PATH)
+    return _load_table(path or TRUE_MAPPINGS_PATH)
 
 
-def append_true_mappings(m: Iterable[Mapping[str, str]], sort: bool = True) -> None:
+def append_true_mappings(
+    m: Iterable[Mapping[str, str]],
+    *,
+    sort: bool = True,
+    path: Optional[Path] = None,
+) -> None:
     """Append new lines to the mappings table."""
-    _write_helper(MAPPINGS_HEADER, m, TRUE_MAPPINGS_PATH, mode="a")
+    if path is None:
+        path = TRUE_MAPPINGS_PATH
+    _write_helper(MAPPINGS_HEADER, m, path, mode="a")
     if sort:
-        lint_true_mappings()
+        lint_true_mappings(path=path)
 
 
 def append_true_mapping_tuples(mappings: Iterable[MappingTuple]) -> None:
     """Append new lines to the mappings table."""
     append_true_mappings(mapping.as_dict() for mapping in set(mappings))
 
 
-def write_true_mappings(m: Iterable[Mapping[str, str]]) -> None:
+def write_true_mappings(m: Iterable[Mapping[str, str]], *, path: Optional[Path] = None) -> None:
     """Write mappigns to the true mappings file."""
-    _write_helper(MAPPINGS_HEADER, m, TRUE_MAPPINGS_PATH, mode="w")
+    _write_helper(MAPPINGS_HEADER, m, path or TRUE_MAPPINGS_PATH, mode="w")
 
 
-def lint_true_mappings(*, standardize: bool = False) -> None:
+def lint_true_mappings(*, standardize: bool = False, path: Optional[Path] = None) -> None:
     """Lint the true mappings file."""
-    mappings = load_mappings()
+    mappings = load_mappings(path=path)
     mappings = _remove_redundant(mappings, MappingTuple, standardize=standardize)
-    write_true_mappings(sorted(mappings, key=mapping_sort_key))
+    write_true_mappings(sorted(mappings, key=mapping_sort_key), path=path)
 
 
 FALSE_MAPPINGS_PATH = get_resource_file_path("incorrect.tsv")
 
 
-def load_false_mappings() -> List[Dict[str, str]]:
+def load_false_mappings(*, path: Optional[Path] = None) -> List[Dict[str, str]]:
     """Load the false mappings table."""
-    return _load_table(FALSE_MAPPINGS_PATH)
+    return _load_table(path or FALSE_MAPPINGS_PATH)
 
 
-def append_false_mappings(m: Iterable[Mapping[str, str]], sort: bool = True) -> None:
+def append_false_mappings(
+    m: Iterable[Mapping[str, str]],
+    *,
+    sort: bool = True,
+    path: Optional[Path] = None,
+) -> None:
     """Append new lines to the false mappings table."""
-    _write_helper(MAPPINGS_HEADER, m, FALSE_MAPPINGS_PATH, mode="a")
+    if path is None:
+        path = FALSE_MAPPINGS_PATH
+    _write_helper(MAPPINGS_HEADER, m, path, mode="a")
     if sort:
-        lint_false_mappings()
+        lint_false_mappings(path=path)
 
 
-def write_false_mappings(m: Iterable[Mapping[str, str]]) -> None:
+def write_false_mappings(m: Iterable[Mapping[str, str]], *, path: Optional[Path] = None) -> None:
     """Write mappings to the false mappings file."""
-    _write_helper(MAPPINGS_HEADER, m, FALSE_MAPPINGS_PATH, mode="w")
+    _write_helper(MAPPINGS_HEADER, m, path or FALSE_MAPPINGS_PATH, mode="w")
 
 
-def lint_false_mappings(*, standardize: bool = False) -> None:
+def lint_false_mappings(*, standardize: bool = False, path: Optional[Path] = None) -> None:
     """Lint the false mappings file."""
-    mappings = load_false_mappings()
+    mappings = load_false_mappings(path=path)
     mappings = _remove_redundant(mappings, MappingTuple, standardize=standardize)
-    write_false_mappings(sorted(mappings, key=mapping_sort_key))
+    write_false_mappings(sorted(mappings, key=mapping_sort_key), path=path)
 
 
 UNSURE_PATH = get_resource_file_path("unsure.tsv")
 
 
-def load_unsure() -> List[Dict[str, str]]:
+def load_unsure(*, path: Optional[Path] = None) -> List[Dict[str, str]]:
     """Load the unsure table."""
-    return _load_table(UNSURE_PATH)
+    return _load_table(path or UNSURE_PATH)
 
 
-def append_unsure_mappings(m: Iterable[Mapping[str, str]], sort: bool = True) -> None:
+def append_unsure_mappings(
+    m: Iterable[Mapping[str, str]],
+    *,
+    sort: bool = True,
+    path: Optional[Path] = None,
+) -> None:
     """Append new lines to the "unsure" mappings table."""
-    _write_helper(MAPPINGS_HEADER, m, UNSURE_PATH, mode="a")
+    if path is None:
+        path = UNSURE_PATH
+    _write_helper(MAPPINGS_HEADER, m, path, mode="a")
     if sort:
-        lint_unsure_mappings()
+        lint_unsure_mappings(path=path)
 
 
-def write_unsure_mappings(m: Iterable[Mapping[str, str]]) -> None:
+def write_unsure_mappings(m: Iterable[Mapping[str, str]], *, path: Optional[Path] = None) -> None:
     """Write mappings to the unsure mappings file."""
-    _write_helper(MAPPINGS_HEADER, m, UNSURE_PATH, mode="w")
+    _write_helper(MAPPINGS_HEADER, m, path or UNSURE_PATH, mode="w")
 
 
-def lint_unsure_mappings(*, standardize: bool = False) -> None:
+def lint_unsure_mappings(*, standardize: bool = False, path: Optional[Path] = None) -> None:
     """Lint the unsure mappings file."""
-    mappings = load_unsure()
+    mappings = load_unsure(path=path)
     mappings = _remove_redundant(mappings, MappingTuple, standardize=standardize)
-    write_unsure_mappings(sorted(mappings, key=mapping_sort_key))
+    write_unsure_mappings(sorted(mappings, key=mapping_sort_key), path=path)
 
 
 PREDICTIONS_PATH = get_resource_file_path("predictions.tsv")
@@ -425,7 +445,7 @@ def load_curators():
 
 
 def filter_predictions(custom_filter: Mapping[str, Mapping[str, Mapping[str, str]]]) -> None:
-    """Filter all of the predictions by removing what's in the custom filter then re-write.
+    """Filter all the predictions by removing what's in the custom filter then re-write.
 
     :param custom_filter: A filter 3-dictionary of source prefix to target prefix
         to source identifier to target identifier

src/biomappings/resources/mappings.tsv

@@ -4313,6 +4313,7 @@ mesh	C000637995	Ruminococcus gnavus	skos:exactMatch	ncit	C124373	Ruminococcus gn
 mesh	C000639838	Sphingomonas paucimobilis	skos:exactMatch	ncit	C86749	Sphingomonas paucimobilis	semapv:ManualMappingCuration	orcid:0000-0003-4423-4370			
 mesh	C000650035	Citrobacter farmeri	skos:exactMatch	ncit	C86264	Citrobacter farmeri	semapv:ManualMappingCuration	orcid:0000-0003-4423-4370			
 mesh	C000654629	H3B-6545	skos:exactMatch	ncit	C142980	Selective Estrogen Receptor Covalent Antagonist H3B-6545	semapv:ManualMappingCuration	orcid:0000-0003-4423-4370			
+mesh	C000655084	red meat allergy	skos:exactMatch	hp	HP:0410319	Alpha-gal allergy	semapv:ManualMappingCuration	orcid:0000-0003-4423-4370	semapv:LexicalMatching	https://github.com/biomappings/biomappings/blob/a80ed2/scripts/import_gilda_mappings.py	0.95
 mesh	C000655246	ACT-774312	skos:exactMatch	ncit	C161834	CRTH2 Antagonist ACT-774312	semapv:ManualMappingCuration	orcid:0000-0003-4423-4370			
 mesh	C000655292	unicuspid aortic valve	skos:exactMatch	hp	HP:0012561	Unicuspid aortic valve	semapv:ManualMappingCuration	orcid:0000-0003-4423-4370			
 mesh	C000656407	MSANTD3 protein, human	skos:exactMatch	uniprot	Q96H12	MSANTD3	semapv:ManualMappingCuration	orcid:0000-0003-4423-4370			
@@ -4335,11 +4336,14 @@ mesh	C000706695	ABBV-744	skos:exactMatch	ncit	C148415	BET Inhibitor ABBV-744	sem
 mesh	C000706947	FGF21 protein, human	skos:exactMatch	uniprot	Q9NSA1	FGF21	semapv:ManualMappingCuration	orcid:0000-0001-9439-5346			
 mesh	C000715747	Telangiectasia macularis eruptiva perstans	skos:exactMatch	hp	HP:0007583	Telangiectasia macularis eruptiva perstans	semapv:ManualMappingCuration	orcid:0000-0001-9439-5346	semapv:LexicalMatching	generate_hp_mesh_mappings.py	0.9
 mesh	C000718787	AA amyloidosis	skos:exactMatch	hp	HP:4000041	AA amyloidosis	semapv:ManualMappingCuration	orcid:0000-0001-9439-5346	semapv:LexicalMatching	generate_hp_mesh_mappings.py	0.9
+mesh	C000718810	anti-amphiphysin autoantibody	skos:exactMatch	hp	HP:5000002	Anti-Amphiphysin antibody	semapv:ManualMappingCuration	orcid:0000-0003-4423-4370	semapv:LexicalMatching	generate_hp_mesh_mappings.py	0.9
 mesh	C000718811	Anti-ARHGAP26 antibody	skos:exactMatch	hp	HP:5000003	Anti-ARHGAP26 antibody	semapv:ManualMappingCuration	orcid:0000-0001-9439-5346	semapv:LexicalMatching	generate_hp_mesh_mappings.py	0.9
 mesh	C000718813	Anti-CARP VIII antibody	skos:exactMatch	hp	HP:5000004	Anti-CARP VIII antibody	semapv:ManualMappingCuration	orcid:0000-0001-9439-5346	semapv:LexicalMatching	generate_hp_mesh_mappings.py	0.9
+mesh	C000719029	anti-GFAP autoantibodies	skos:exactMatch	hp	HP:5000013	Anti-GFAP antibody	semapv:ManualMappingCuration	orcid:0000-0003-4423-4370	semapv:LexicalMatching	generate_hp_mesh_mappings.py	0.9
 mesh	C000719042	anti-ryanodine receptor autoantibody	skos:exactMatch	hp	HP:5000047	Anti-ryanodine receptor antibody	semapv:ManualMappingCuration	orcid:0000-0003-4423-4370	semapv:LexicalMatching	generate_hp_mesh_mappings.py	0.9
 mesh	C000719407	Midline brainstem cleft	skos:exactMatch	hp	HP:0033645	Midline brainstem cleft	semapv:ManualMappingCuration	orcid:0000-0001-9439-5346	semapv:LexicalMatching	generate_hp_mesh_mappings.py	0.9
 mesh	C000719531	misophonia	skos:exactMatch	hp	HP:0025113	Misophonia	semapv:ManualMappingCuration	orcid:0000-0001-9439-5346	semapv:LexicalMatching	generate_hp_mesh_mappings.py	0.9
+mesh	C000721288	Cleft eyelid	skos:exactMatch	hp	HP:0000625	Eyelid coloboma	semapv:ManualMappingCuration	orcid:0000-0003-4423-4370	semapv:LexicalMatching	generate_hp_mesh_mappings.py	0.9
 mesh	C000721322	Coarse facial features	skos:exactMatch	hp	HP:0000280	Coarse facial features	semapv:ManualMappingCuration	orcid:0000-0001-9439-5346	semapv:LexicalMatching	generate_hp_mesh_mappings.py	0.9
 mesh	C000721349	Air crescent sign	skos:exactMatch	hp	HP:0032172	Air crescent sign	semapv:ManualMappingCuration	orcid:0000-0001-9439-5346	semapv:LexicalMatching	generate_hp_mesh_mappings.py	0.9
 mesh	C000721350	Silhouette sign	skos:exactMatch	hp	HP:0033647	Silhouette sign	semapv:ManualMappingCuration	orcid:0000-0001-9439-5346	semapv:LexicalMatching	generate_hp_mesh_mappings.py	0.9

src/biomappings/resources/predictions.tsv

@@ -14130,7 +14130,6 @@ mesh	C000655028	CR2 protein, human	skos:exactMatch	uniprot	P20023	CR2	semapv:Lex
 mesh	C000655029	ABHD15 protein, human	skos:exactMatch	uniprot	Q6UXT9	ABHD15	semapv:LexicalMatching	0.999	https://github.com/biomappings/biomappings/blob/a840cf/scripts/generate_mesh_uniprot_mappings.py
 mesh	C000655039	ME3 protein, human	skos:exactMatch	uniprot	Q16798	ME3	semapv:LexicalMatching	0.999	https://github.com/biomappings/biomappings/blob/a840cf/scripts/generate_mesh_uniprot_mappings.py
 mesh	C000655044	BCHE protein, human	skos:exactMatch	uniprot	P06276	BCHE	semapv:LexicalMatching	0.999	https://github.com/biomappings/biomappings/blob/a840cf/scripts/generate_mesh_uniprot_mappings.py
-mesh	C000655084	red meat allergy	skos:exactMatch	hp	HP:0410319	Alpha-gal allergy	semapv:LexicalMatching	0.95	https://github.com/biomappings/biomappings/blob/a80ed2/scripts/import_gilda_mappings.py
 mesh	C000655129	MINDY4 protein, human	skos:exactMatch	uniprot	Q4G0A6	MINDY4	semapv:LexicalMatching	0.999	https://github.com/biomappings/biomappings/blob/a840cf/scripts/generate_mesh_uniprot_mappings.py
 mesh	C000655132	SHISA3 protein, human	skos:exactMatch	uniprot	A0PJX4	SHISA3	semapv:LexicalMatching	0.999	https://github.com/biomappings/biomappings/blob/a840cf/scripts/generate_mesh_uniprot_mappings.py
 mesh	C000655134	TXNDC9 protein, human	skos:exactMatch	uniprot	O14530	TXNDC9	semapv:LexicalMatching	0.999	https://github.com/biomappings/biomappings/blob/a840cf/scripts/generate_mesh_uniprot_mappings.py
@@ -14291,10 +14290,7 @@ mesh	C000707228	MYMX protein, human	skos:exactMatch	uniprot	A0A1B0GTQ4	MYMX	sema
 mesh	C000707256	AAMDC protein, human	skos:exactMatch	uniprot	Q9H7C9	AAMDC	semapv:LexicalMatching	0.999	https://github.com/biomappings/biomappings/blob/a840cf/scripts/generate_mesh_uniprot_mappings.py
 mesh	C000707475	HTR2A protein, human	skos:exactMatch	uniprot	P28223	HTR2A	semapv:LexicalMatching	0.999	https://github.com/biomappings/biomappings/blob/a840cf/scripts/generate_mesh_uniprot_mappings.py
 mesh	C000707567	CG428	skos:exactMatch	ncit	C125063	Botanical Lotion CG428	semapv:LexicalMatching	0.95	https://github.com/biomappings/biomappings/blob/a80ed2/scripts/import_gilda_mappings.py
-mesh	C000718810	anti-amphiphysin autoantibody	skos:exactMatch	hp	HP:5000002	Anti-Amphiphysin antibody	semapv:LexicalMatching	0.9	generate_hp_mesh_mappings.py
-mesh	C000719029	anti-GFAP autoantibodies	skos:exactMatch	hp	HP:5000013	Anti-GFAP antibody	semapv:LexicalMatching	0.9	generate_hp_mesh_mappings.py
 mesh	C000721268	Isolated left subclavian artery	skos:exactMatch	hp	HP:0031633	Isolation of the left subclavian artery	semapv:LexicalMatching	0.9	generate_hp_mesh_mappings.py
-mesh	C000721288	Cleft eyelid	skos:exactMatch	hp	HP:0000625	Eyelid coloboma	semapv:LexicalMatching	0.9	generate_hp_mesh_mappings.py
 mesh	C000721289	Malar hypoplasia	skos:exactMatch	hp	HP:0000272	Malar flattening	semapv:LexicalMatching	0.9	generate_hp_mesh_mappings.py
 mesh	C000721389	tram track sign optic nerve	skos:exactMatch	hp	HP:0032270	Optic nerve tram-track sign	semapv:LexicalMatching	0.9	generate_hp_mesh_mappings.py
 mesh	C000847	daunorubicinol	skos:exactMatch	ncit	C1062	Duborimycin	semapv:LexicalMatching	0.95	https://github.com/biomappings/biomappings/blob/a80ed2/scripts/import_gilda_mappings.py

src/biomappings/templates/summary.html

@@ -10,7 +10,7 @@
         {{ util.render_messages(dismissible=True, container=False) }}
         <div class="row">
             <div class="card">
-                <h5 class="card-header text-center">Biomappings Su</h5>
+                <h5 class="card-header text-center">Biomappings Summary</h5>
                 <div class="card-body">
                     <p>Summary text.</p>
                 </div>

src/biomappings/utils.py

@@ -4,16 +4,18 @@
 
 import os
 import re
+from pathlib import Path
 from subprocess import CalledProcessError, check_output  # noqa: S404
 from typing import Any, Mapping, Optional, Tuple
 
 import bioregistry
 
-HERE = os.path.dirname(os.path.abspath(__file__))
-RESOURCE_PATH = os.path.abspath(os.path.join(HERE, "resources"))
-DOCS = os.path.abspath(os.path.join(HERE, os.pardir, os.pardir, "docs"))
-IMG = os.path.join(DOCS, "img")
-DATA = os.path.join(DOCS, "_data")
+HERE = Path(__file__).parent.resolve()
+ROOT = HERE.parent.parent.resolve()
+RESOURCE_PATH = HERE.joinpath("resources")
+DOCS = ROOT.joinpath("docs")
+IMG = DOCS.joinpath("img")
+DATA = DOCS.joinpath("_data")
 
 OVERRIDE_MIRIAM = {
     # ITO is very messy (combines mostly numbers with a few

src/biomappings/wsgi.py

@@ -100,14 +100,23 @@ def url_for_state(endpoint, state: State, **kwargs) -> str:
 def get_app(
     target_curies: Optional[Iterable[Tuple[str, str]]] = None,
     predictions_path: Optional[Path] = None,
+    positives_path: Optional[Path] = None,
+    negatives_path: Optional[Path] = None,
+    unsure_path: Optional[Path] = None,
 ) -> flask.Flask:
     """Get a curation flask app."""
     app_ = flask.Flask(__name__)
     app_.config["WTF_CSRF_ENABLED"] = False
     app_.config["SECRET_KEY"] = os.urandom(8)
     app_.config["SHOW_RELATIONS"] = True
     app_.config["SHOW_LINES"] = False
-    controller = Controller(target_curies=target_curies, predictions_path=predictions_path)
+    controller = Controller(
+        target_curies=target_curies,
+        predictions_path=predictions_path,
+        positives_path=positives_path,
+        negatives_path=negatives_path,
+        unsure_path=unsure_path,
+    )
     app_.config["controller"] = controller
     flask_bootstrap.Bootstrap4(app_)
     app_.register_blueprint(blueprint)
@@ -137,16 +146,27 @@ def __init__(
         *,
         target_curies: Optional[Iterable[Tuple[str, str]]] = None,
         predictions_path: Optional[Path] = None,
+        positives_path: Optional[Path] = None,
+        negatives_path: Optional[Path] = None,
+        unsure_path: Optional[Path] = None,
     ):
         """Instantiate the web controller.
 
         :param target_curies: Pairs of prefix, local unique identifiers that are the target
             of curation. If this is given, pre-filters will be made before on predictions
             to only show ones where either the source or target appears in this set
         :param predictions_path: A custom predictions file to curate from
+        :param positives_path: A custom positives file to curate to
+        :param negatives_path: A custom negatives file to curate to
+        :param unsure_path: A custom unsure file to curate to
         """
         self.predictions_path = predictions_path
         self._predictions = load_predictions(path=self.predictions_path)
+
+        self.positives_path = positives_path
+        self.negatives_path = negatives_path
+        self.unsure_path = unsure_path
+
         self._marked: Dict[int, str] = {}
         self.total_curated = 0
         self._added_mappings: List[Dict[str, Union[None, str, float]]] = []
@@ -442,14 +462,14 @@ def persist(self):
             prediction["type"] = "semapv:ManualMappingCuration"
             entries[value].append(prediction)
 
-        append_true_mappings(entries["correct"])
-        append_false_mappings(entries["incorrect"])
-        append_unsure_mappings(entries["unsure"])
+        append_true_mappings(entries["correct"], path=self.positives_path)
+        append_false_mappings(entries["incorrect"], path=self.negatives_path)
+        append_unsure_mappings(entries["unsure"], path=self.unsure_path)
         write_predictions(self._predictions, path=self.predictions_path)
         self._marked.clear()
 
         # Now add manually curated mappings
-        append_true_mappings(self._added_mappings)
+        append_true_mappings(self._added_mappings, path=self.positives_path)
         self._added_mappings = []