sc-graft 0.0.0__tar.gz

sc_graft-0.0.0/PKG-INFO

@@ -0,0 +1,39 @@
+Metadata-Version: 2.4
+Name: sc-graft
+Version: 0.0.0
+Summary: single-cell Graph of Receptors, pAthways, Factors and Targets — external-prior loaders (early placeholder; core model not yet released)
+Author-email: Ohbin Kwon <ohbin.kwon01@gmail.com>
+License: MIT
+Project-URL: Homepage, https://github.com/ObKwon115/sc-graft
+Project-URL: Repository, https://github.com/ObKwon115/sc-graft
+Project-URL: Issues, https://github.com/ObKwon115/sc-graft/issues
+Keywords: single-cell,scRNA-seq,graph-neural-network,cell-signaling,bioinformatics
+Classifier: Development Status :: 1 - Planning
+Classifier: Intended Audience :: Science/Research
+Classifier: License :: OSI Approved :: MIT License
+Classifier: Operating System :: OS Independent
+Classifier: Programming Language :: Python :: 3
+Classifier: Topic :: Scientific/Engineering :: Bio-Informatics
+Requires-Python: >=3.9
+Description-Content-Type: text/markdown
+Provides-Extra: scenic
+Requires-Dist: pandas>=1.5; extra == "scenic"
+
+# sc-graft
+
+**Early placeholder release.** `sc-graft` — a single-cell **G**raph of
+**R**eceptors, p**A**thways, **F**actors and **T**argets — is a
+biologically-grounded heterogeneous graph transformer for inferring how receptor
+signaling propagates to target genes in single cells.
+
+This pre-release ships only a small slice of the **preprocessing** layer: the
+**external-prior loaders** (STRING physical PPI, MSigDB gene sets, SCENIC
+regulons). The network-construction logic and the model itself are under active
+development and are **not included yet**.
+
+```python
+from sc_graft import load_string_interactions, read_gmt, gmt_path, load_regulons
+```
+
+- Source / docs: https://github.com/ObKwon115/sc-graft
+- License: MIT

sc_graft-0.0.0/README.md

@@ -0,0 +1,18 @@
+# sc-graft
+
+**Early placeholder release.** `sc-graft` — a single-cell **G**raph of
+**R**eceptors, p**A**thways, **F**actors and **T**argets — is a
+biologically-grounded heterogeneous graph transformer for inferring how receptor
+signaling propagates to target genes in single cells.
+
+This pre-release ships only a small slice of the **preprocessing** layer: the
+**external-prior loaders** (STRING physical PPI, MSigDB gene sets, SCENIC
+regulons). The network-construction logic and the model itself are under active
+development and are **not included yet**.
+
+```python
+from sc_graft import load_string_interactions, read_gmt, gmt_path, load_regulons
+```
+
+- Source / docs: https://github.com/ObKwon115/sc-graft
+- License: MIT

sc_graft-0.0.0/pyproject.toml

@@ -0,0 +1,33 @@
+[build-system]
+requires = ["setuptools>=68", "wheel"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "sc-graft"
+version = "0.0.0"
+description = "single-cell Graph of Receptors, pAthways, Factors and Targets — external-prior loaders (early placeholder; core model not yet released)"
+readme = "README.md"
+requires-python = ">=3.9"
+license = { text = "MIT" }
+authors = [{ name = "Ohbin Kwon", email = "ohbin.kwon01@gmail.com" }]
+keywords = ["single-cell", "scRNA-seq", "graph-neural-network", "cell-signaling", "bioinformatics"]
+classifiers = [
+  "Development Status :: 1 - Planning",
+  "Intended Audience :: Science/Research",
+  "License :: OSI Approved :: MIT License",
+  "Operating System :: OS Independent",
+  "Programming Language :: Python :: 3",
+  "Topic :: Scientific/Engineering :: Bio-Informatics",
+]
+dependencies = []
+
+[project.optional-dependencies]
+scenic = ["pandas>=1.5"]
+
+[project.urls]
+Homepage = "https://github.com/ObKwon115/sc-graft"
+Repository = "https://github.com/ObKwon115/sc-graft"
+Issues = "https://github.com/ObKwon115/sc-graft/issues"
+
+[tool.setuptools.packages.find]
+where = ["src"]

sc_graft-0.0.0/setup.cfg

@@ -0,0 +1,4 @@
+[egg_info]
+tag_build = 
+tag_date = 0
+

sc_graft-0.0.0/src/sc_graft/__init__.py

@@ -0,0 +1,30 @@
+"""sc-graft — single-cell Graph of Receptors, pAthways, Factors and Targets.
+
+Early placeholder. Only the external-prior loaders are published so far; the
+network-construction logic and the model are not yet released.
+See https://github.com/ObKwon115/sc-graft.
+"""
+from .priors import (
+    TAXID,
+    MSIGDB_SUFFIX,
+    MSIGDB_STEM,
+    string_paths,
+    load_string_interactions,
+    gmt_path,
+    read_gmt,
+    load_regulons,
+)
+
+__version__ = "0.0.0"
+
+__all__ = [
+    "TAXID",
+    "MSIGDB_SUFFIX",
+    "MSIGDB_STEM",
+    "string_paths",
+    "load_string_interactions",
+    "gmt_path",
+    "read_gmt",
+    "load_regulons",
+    "__version__",
+]

sc_graft-0.0.0/src/sc_graft/priors.py

@@ -0,0 +1,124 @@
+"""External prior loaders (species-aware).
+
+All edge priors come from public databases; this module resolves the right
+files for mouse or human and loads them. Nothing here is dataset-specific.
+
+Resources:
+  STRING   physical PPI            data/prior/STRING/<taxid>.protein.*.v12.0.txt.gz
+  MSigDB   pathway gene sets       data/prior/misgdb/*.<Mm|Hs>.symbols.gmt
+  SCENIC   regulons (your run)     <proc>/scenic/regulons.csv
+"""
+from __future__ import annotations
+import gzip
+from collections import defaultdict
+from pathlib import Path
+from typing import Dict, List, Set
+
+TAXID = {"mouse": "10090", "human": "9606"}
+MSIGDB_SUFFIX = {"mouse": "Mm", "human": "Hs"}
+
+
+# --------------------------------------------------------------------------- #
+# STRING physical PPI
+# --------------------------------------------------------------------------- #
+def string_paths(species: str, prior_root: str | Path) -> Dict[str, Path]:
+    """Resolve the three STRING files for a species."""
+    tax = TAXID[species]
+    root = Path(prior_root) / "STRING"
+    return {
+        "info": root / f"{tax}.protein.info.v12.0.txt.gz",
+        "links": root / f"{tax}.protein.links.v12.0.txt.gz",
+        "physical": root / f"{tax}.protein.physical.links.v12.0.txt.gz",
+    }
+
+
+def load_string_interactions(
+    species: str,
+    prior_root: str | Path,
+    *,
+    gene_universe: Set[str],
+    score_thr: int = 700,
+    physical: bool = True,
+) -> Dict[str, Set[str]]:
+    """Gene -> set of genes it interacts with (within ``gene_universe``).
+
+    ``physical=True`` uses the physical-binding subnetwork (expression-
+    independent); ``False`` uses the combined score (includes coexpression).
+    """
+    paths = string_paths(species, prior_root)
+    info, links = paths["info"], paths["physical"] if physical else paths["links"]
+    for p in (info, links):
+        if not p.exists():
+            raise FileNotFoundError(
+                f"STRING file missing: {p}\n  download the STRING protein.info and "
+                f"protein.physical.links files for taxid {TAXID[species]} from "
+                f"https://string-db.org/cgi/download"
+            )
+    # protein id <-> gene symbol
+    name2prot: Dict[str, Set[str]] = defaultdict(set)
+    prot2name: Dict[str, str] = {}
+    with gzip.open(info, "rt") as f:
+        next(f)
+        for ln in f:
+            x = ln.split("\t")
+            prot2name[x[0]] = x[1]
+            name2prot[x[1]].add(x[0])
+    keep = {pr for g in gene_universe for pr in name2prot.get(g, ())}
+    inter: Dict[str, Set[str]] = defaultdict(set)
+    with gzip.open(links, "rt") as f:
+        next(f)
+        for ln in f:
+            p1, p2, s = ln.split()
+            if int(s) < score_thr:
+                continue
+            if p1 in keep and p2 in keep:
+                a, b = prot2name[p1], prot2name[p2]
+                inter[a].add(b)
+                inter[b].add(a)
+    return inter
+
+
+# --------------------------------------------------------------------------- #
+# MSigDB gene sets (WikiPathways / Reactome / Hallmark)
+# --------------------------------------------------------------------------- #
+# MSigDB collection names differ by species: mouse uses m2/mh/m5, human c2/h/c5.
+MSIGDB_STEM = {
+    "mouse": {"wikipathways": "m2.cp.wikipathways", "reactome": "m2.cp.reactome",
+              "hallmark": "mh.all", "gobp": "m5.go.bp"},
+    "human": {"wikipathways": "c2.cp.wikipathways", "reactome": "c2.cp.reactome",
+              "hallmark": "h.all", "gobp": "c5.go.bp"},
+}
+
+
+def gmt_path(species: str, prior_root: str | Path, collection: str = "wikipathways",
+             version: str = "2026.1") -> Path:
+    """e.g. mouse m2.cp.wikipathways.v2026.1.Mm.symbols.gmt
+            human c2.cp.wikipathways.v2026.1.Hs.symbols.gmt"""
+    sfx = MSIGDB_SUFFIX[species]
+    stem = MSIGDB_STEM[species][collection]
+    return Path(prior_root) / "misgdb" / f"{stem}.v{version}.{sfx}.symbols.gmt"
+
+
+def read_gmt(path: str | Path) -> Dict[str, List[str]]:
+    path = Path(path)
+    if not path.exists():
+        raise FileNotFoundError(f"GMT missing: {path}")
+    out: Dict[str, List[str]] = {}
+    for ln in open(path):
+        x = ln.rstrip("\n").split("\t")
+        out[x[0]] = [g for g in x[2:] if g]
+    return out
+
+
+# --------------------------------------------------------------------------- #
+# SCENIC regulons (from the user's own SCENIC run)
+# --------------------------------------------------------------------------- #
+def load_regulons(path: str | Path):
+    """SCENIC regulons table with columns TF, target[, weight]."""
+    import pandas as pd
+    path = Path(path)
+    if not path.exists():
+        raise FileNotFoundError(
+            f"SCENIC regulons missing: {path}\n  run pySCENIC on your counts first"
+        )
+    return pd.read_csv(path)

sc_graft-0.0.0/src/sc_graft.egg-info/PKG-INFO

@@ -0,0 +1,39 @@
+Metadata-Version: 2.4
+Name: sc-graft
+Version: 0.0.0
+Summary: single-cell Graph of Receptors, pAthways, Factors and Targets — external-prior loaders (early placeholder; core model not yet released)
+Author-email: Ohbin Kwon <ohbin.kwon01@gmail.com>
+License: MIT
+Project-URL: Homepage, https://github.com/ObKwon115/sc-graft
+Project-URL: Repository, https://github.com/ObKwon115/sc-graft
+Project-URL: Issues, https://github.com/ObKwon115/sc-graft/issues
+Keywords: single-cell,scRNA-seq,graph-neural-network,cell-signaling,bioinformatics
+Classifier: Development Status :: 1 - Planning
+Classifier: Intended Audience :: Science/Research
+Classifier: License :: OSI Approved :: MIT License
+Classifier: Operating System :: OS Independent
+Classifier: Programming Language :: Python :: 3
+Classifier: Topic :: Scientific/Engineering :: Bio-Informatics
+Requires-Python: >=3.9
+Description-Content-Type: text/markdown
+Provides-Extra: scenic
+Requires-Dist: pandas>=1.5; extra == "scenic"
+
+# sc-graft
+
+**Early placeholder release.** `sc-graft` — a single-cell **G**raph of
+**R**eceptors, p**A**thways, **F**actors and **T**argets — is a
+biologically-grounded heterogeneous graph transformer for inferring how receptor
+signaling propagates to target genes in single cells.
+
+This pre-release ships only a small slice of the **preprocessing** layer: the
+**external-prior loaders** (STRING physical PPI, MSigDB gene sets, SCENIC
+regulons). The network-construction logic and the model itself are under active
+development and are **not included yet**.
+
+```python
+from sc_graft import load_string_interactions, read_gmt, gmt_path, load_regulons
+```
+
+- Source / docs: https://github.com/ObKwon115/sc-graft
+- License: MIT

sc_graft-0.0.0/src/sc_graft.egg-info/SOURCES.txt

@@ -0,0 +1,9 @@
+README.md
+pyproject.toml
+src/sc_graft/__init__.py
+src/sc_graft/priors.py
+src/sc_graft.egg-info/PKG-INFO
+src/sc_graft.egg-info/SOURCES.txt
+src/sc_graft.egg-info/dependency_links.txt
+src/sc_graft.egg-info/requires.txt
+src/sc_graft.egg-info/top_level.txt

sc_graft-0.0.0/src/sc_graft.egg-info/dependency_links.txt

@@ -0,0 +1 @@
+

sc_graft-0.0.0/src/sc_graft.egg-info/requires.txt

@@ -0,0 +1,3 @@
+
+[scenic]
+pandas>=1.5

sc_graft-0.0.0/src/sc_graft.egg-info/top_level.txt

@@ -0,0 +1 @@
+sc_graft