setup pipeline for multi-configuration runs

src/depiction_targeted_preproc/example/run.py

@@ -15,12 +15,13 @@
         "qc/plot_marker_presence.pdf",
         "qc/plot_peak_density_combined.pdf",
         "qc/plot_peak_density_grouped.pdf",
-        "qc/plot_calibration_map.pdf",
     ],
     PipelineArtifact.CALIB_IMAGES: ["images_default.ome.tiff", "images_default_norm.ome.tiff"],
     # PipelineArtifact.CALIB_HEATMAP: ["images_calib_heatmap.ome.tiff"],
     # TODO
-    PipelineArtifact.CALIB_HEATMAP: [],
+    PipelineArtifact.CALIB_HEATMAP: [
+        "qc/plot_calibration_map.pdf",
+    ],
     PipelineArtifact.DEBUG: [
         "qc/plot_marker_presence_cv.pdf",
         "qc/plot_spectra_for_marker.pdf",
@@ -122,7 +123,7 @@ def snakemake_invoke(work_dir: Path, result_files: list[Path]) -> None:
         "1",
         "--snakefile",
         str(snakefile_path),
-        *[str(file) for file in result_files],
+        *[str(file.relative_to(work_dir)) for file in result_files],
     ]
     logger.info("Executing {command}", command=command)
     subprocess.run(

src/depiction_targeted_preproc/example_compare/run_compare.py

@@ -10,14 +10,26 @@ def get_configs() -> dict[str, Path]:
     return {path.stem: path for path in config_dir.glob("*.yml")}
 
 
-def main() -> None:
-    data_raw_path = Path(__file__).parent / "data-raw"
-    input_imzml = data_raw_path / "menzha_20231208_s607923_tonsil-repro-sample-01.imzML"
-    input_mass_list = data_raw_path / "mass_list_vend.csv"
+def prepare_tasks(input_imzml_path: Path, work_dir: Path) -> list[Path]:
+    input_mass_list = input_imzml_path.parent / "mass_list_vend.csv"
+    folders = set_up_work_dir(work_dir, input_imzml_path, input_mass_list)
+    requested_files = get_all_output_files(folders)
+
+    exp_files = [work_dir / input_imzml_path.stem / "exp_compare_cluster_stats.pdf"]
+    return requested_files + exp_files
+
 
+def main() -> None:
     work_dir = Path(__file__).parent / "data-work"
-    folders = set_up_work_dir(work_dir, input_imzml, input_mass_list)
-    requested_files = get_all_output_files(folders)
+    data_raw_dir = Path(__file__).parent / "data-raw"
+
+    requested_files = []
+    requested_files += prepare_tasks(
+        data_raw_dir / "menzha_20231208_s607923_tonsil-repro-sample-01.imzML", work_dir=work_dir
+    )
+    requested_files += prepare_tasks(
+        data_raw_dir / "menzha_20231208_s607930_64074-b20-30928-a.imzML", work_dir=work_dir
+    )
 
     snakemake_invoke(work_dir=work_dir, result_files=requested_files)
 
@@ -37,8 +49,9 @@ def get_all_output_files(folders: list[Path]) -> list[Path]:
 def set_up_work_dir(work_dir: Path, input_imzml: Path, input_mass_list: Path) -> list[Path]:
     configs = get_configs()
     folders = []
+    sample_name = input_imzml.stem
     for config_name, config_path in configs.items():
-        dir = work_dir / config_name
+        dir = work_dir / sample_name / config_name
         initial_setup(input_imzml=input_imzml, input_mass_list=input_mass_list, params_file=config_path, dir=dir)
         folders.append(dir)
     return folders

src/depiction_targeted_preproc/workflow/exp/compare_cluster_stats.py

@@ -0,0 +1,25 @@
+from pathlib import Path
+from typing import Annotated
+
+import polars as pl
+import typer
+from typer import Option
+
+
+def load_data(csv_paths: list[Path]) -> pl.DataFrame:
+    collect = []
+    for csv_path in csv_paths:
+        df = pl.read_csv(csv_path)
+        collect.append(df.with_columns(label=pl.lit(csv_path.parent.name)))
+    return pl.concat(collect)
+
+
+def compare_cluster_stats(input_csv_path: list[Path], output_pdf: Annotated[Path, Option()]) -> None:
+    data = load_data(input_csv_path)
+    print(data)
+
+
+if __name__ == "__main__":
+    typer.run(compare_cluster_stats)
+
+# print(sys.argv)

src/depiction_targeted_preproc/workflow/experimental.smk

@@ -3,3 +3,16 @@ version: "3"
 include: "rules/rules_proc.smk"
 include: "rules/rules_vis.smk"
 include: "rules/rules_qc.smk"
+
+
+exp_variants = ["chem_noise", "mass_cluster", "reg_shift"]
+
+rule exp_compare_cluster_stats:
+    input:
+        csv=expand("{{sample}}/{exp_variant}/cluster_default_stats_kmeans.csv", exp_variant=exp_variants)
+    output:
+        pdf="{sample}/exp_compare_cluster_stats.pdf"
+    shell:
+        "python -m depiction_targeted_preproc.workflow.exp.compare_cluster_stats"
+        " {input.csv}"
+        " --output-pdf {output}"

src/depiction_targeted_preproc/workflow/proc/cluster_kmeans.py

@@ -26,7 +26,7 @@ def retain_strongest_signals(data: xarray.DataArray, n_features: int) -> xarray.
 def cluster_kmeans(input_netcdf_path: Annotated[Path, Option()], output_netcdf_path: Annotated[Path, Option()]) -> None:
     image = MultiChannelImage.read_hdf5(input_netcdf_path)
     # TODO make configurable
-    n_clusters = 10
+    n_clusters = 6
     n_features = 30
 
     reduced_data = retain_strongest_signals(image.data_flat.transpose("i", "c"), n_features)