Merge pull request #77 from daniel-unyi-42/main

Update tutorial notebook

README.md

@@ -2,11 +2,8 @@
 
 [![pre-commit.ci status](https://results.pre-commit.ci/badge/github/EliHei2/segger_dev/main.svg)](https://results.pre-commit.ci/latest/github/EliHei2/segger_dev/main)
 
-
-
 **Important note (Dec 2024)**: As segger is currently undergoing constant development we highly recommending installing segger directly via github.
 
-
 **segger** is a cutting-edge tool for **cell segmentation** in **single-molecule spatial omics** datasets. By leveraging **graph neural networks (GNNs)** and heterogeneous graphs, segger offers unmatched accuracy and scalability.
 
 # How segger Works
@@ -52,7 +49,7 @@ segger tackles these with a **graph-based approach**, achieving superior segment
 
 ---
 
-## Installation 
+## Installation
 
 **Important note (Dec 2024)**: As segger is currently undergoing constant development we highly recommending installing segger directly via github.
 
@@ -78,7 +75,6 @@ pip install pyg_lib torch_scatter torch_sparse torch_cluster torch_spline_conv -
 
 Afterwards choose the installation method that best suits your needs.
 
-
 ### GitHub Installation
 
 For a straightforward local installation from GitHub, clone the repository and install the package using `pip`:

scripts/create_data_fast_sample.py

@@ -7,87 +7,77 @@
 import numpy as np
 from segger.data.parquet._utils import get_polygons_from_xy
 
-xenium_data_dir = Path('data_raw/breast_cancer/Xenium_FFPE_Human_Breast_Cancer_Rep1/outs/')
-segger_data_dir = Path('data_tidy/pyg_datasets/bc_rep1_emb')
+xenium_data_dir = Path("data_raw/breast_cancer/Xenium_FFPE_Human_Breast_Cancer_Rep1/outs/")
+segger_data_dir = Path("data_tidy/pyg_datasets/bc_rep1_emb")
 
 
-scrnaseq_file = Path('/omics/groups/OE0606/internal/tangy/tasks/schier/data/atals_filtered.h5ad')
-celltype_column = 'celltype_major'
-gene_celltype_abundance_embedding = calculate_gene_celltype_abundance_embedding(
-    sc.read(scrnaseq_file),
-    celltype_column
-)
+scrnaseq_file = Path("/omics/groups/OE0606/internal/tangy/tasks/schier/data/atals_filtered.h5ad")
+celltype_column = "celltype_major"
+gene_celltype_abundance_embedding = calculate_gene_celltype_abundance_embedding(sc.read(scrnaseq_file), celltype_column)
 
 sample = STSampleParquet(
     base_dir=xenium_data_dir,
     n_workers=4,
-    sample_type='xenium',
-    weights=gene_celltype_abundance_embedding, # uncomment if gene-celltype embeddings are available
+    sample_type="xenium",
+    weights=gene_celltype_abundance_embedding,  # uncomment if gene-celltype embeddings are available
 )
 
-transcripts = pd.read_parquet(
-    xenium_data_dir / 'transcripts.parquet',
-    filters=[[('overlaps_nucleus', '=', 1)]]
-)
-boundaries = pd.read_parquet(xenium_data_dir / 'nucleus_boundaries.parquet')
+transcripts = pd.read_parquet(xenium_data_dir / "transcripts.parquet", filters=[[("overlaps_nucleus", "=", 1)]])
+boundaries = pd.read_parquet(xenium_data_dir / "nucleus_boundaries.parquet")
 
-sizes = transcripts.groupby('cell_id').size()
-polygons = get_polygons_from_xy(boundaries, 'vertex_x', 'vertex_y', 'cell_id')
+sizes = transcripts.groupby("cell_id").size()
+polygons = get_polygons_from_xy(boundaries, "vertex_x", "vertex_y", "cell_id")
 densities = polygons[sizes.index].area / sizes
 bd_width = polygons.minimum_bounding_radius().median() * 2
 
 # 1/4 median boundary diameter
 dist_tx = bd_width / 4
 # 90th percentile density of bounding circle with radius=dist_tx
-k_tx = math.ceil(np.quantile(dist_tx ** 2 * np.pi * densities, 0.9)) 
+k_tx = math.ceil(np.quantile(dist_tx**2 * np.pi * densities, 0.9))
 
 print(k_tx)
 print(dist_tx)
 
 
 sample.save(
-      data_dir=segger_data_dir,
-      k_bd=3,
-      dist_bd=15.0,
-      k_tx=dist_tx,
-      dist_tx=k_tx,
-      tile_width=120,
-      tile_height=120,
-      neg_sampling_ratio=5.0,
-      frac=1.0,
-      val_prob=0.1,
-      test_prob=0.1,
+    data_dir=segger_data_dir,
+    k_bd=3,
+    dist_bd=15.0,
+    k_tx=dist_tx,
+    dist_tx=k_tx,
+    tile_width=120,
+    tile_height=120,
+    neg_sampling_ratio=5.0,
+    frac=1.0,
+    val_prob=0.1,
+    test_prob=0.1,
 )
 
 
-xenium_data_dir = Path('data_tidy/bc_5k')
-segger_data_dir = Path('data_tidy/pyg_datasets/bc_5k_emb')
-
+xenium_data_dir = Path("data_tidy/bc_5k")
+segger_data_dir = Path("data_tidy/pyg_datasets/bc_5k_emb")
 
 
 sample = STSampleParquet(
     base_dir=xenium_data_dir,
     n_workers=1,
-    sample_type='xenium',
-    weights=gene_celltype_abundance_embedding, # uncomment if gene-celltype embeddings are available
+    sample_type="xenium",
+    weights=gene_celltype_abundance_embedding,  # uncomment if gene-celltype embeddings are available
 )
 
 
-transcripts = pd.read_parquet(
-    xenium_data_dir / 'transcripts.parquet',
-    filters=[[('overlaps_nucleus', '=', 1)]]
-)
-boundaries = pd.read_parquet(xenium_data_dir / 'nucleus_boundaries.parquet')
+transcripts = pd.read_parquet(xenium_data_dir / "transcripts.parquet", filters=[[("overlaps_nucleus", "=", 1)]])
+boundaries = pd.read_parquet(xenium_data_dir / "nucleus_boundaries.parquet")
 
-sizes = transcripts.groupby('cell_id').size()
-polygons = get_polygons_from_xy(boundaries, 'vertex_x', 'vertex_y', 'cell_id')
+sizes = transcripts.groupby("cell_id").size()
+polygons = get_polygons_from_xy(boundaries, "vertex_x", "vertex_y", "cell_id")
 densities = polygons[sizes.index].area / sizes
 bd_width = polygons.minimum_bounding_radius().median() * 2
 
 # 1/4 median boundary diameter
 dist_tx = bd_width / 4
 # 90th percentile density of bounding circle with radius=dist_tx
-k_tx = math.ceil(np.quantile(dist_tx ** 2 * np.pi * densities, 0.9)) 
+k_tx = math.ceil(np.quantile(dist_tx**2 * np.pi * densities, 0.9))
 
 print(k_tx)
 print(dist_tx)

scripts/predict_model_sample.py

@@ -16,12 +16,11 @@
 import dask.dataframe as dd
 
 
-
 seg_tag = "bc_fast_data_emb_major"
 model_version = 1
 
-segger_data_dir = Path('data_tidy/pyg_datasets') / seg_tag
-models_dir = Path("./models") / seg_tag 
+segger_data_dir = Path("data_tidy/pyg_datasets") / seg_tag
+models_dir = Path("./models") / seg_tag
 benchmarks_dir = Path("/dkfz/cluster/gpu/data/OE0606/elihei/segger_experiments/data_tidy/benchmarks/xe_rep1_bc")
 transcripts_file = "data_raw/xenium/Xenium_FFPE_Human_Breast_Cancer_Rep1/transcripts.parquet"
 # Initialize the Lightning data module
@@ -58,4 +57,3 @@
     gpu_ids=["0"],
     # client=client
 )
-