Method: spaTrack

CHANGELOG.md

@@ -6,6 +6,7 @@
 
 ## NEW FUNCTIONALITY
 
+* Added `spaTrack` method (PR #4).
 
 ## MAJOR CHANGES
 

src/methods/spatrack/config.vsh.yaml

@@ -0,0 +1,74 @@
+__merge__: ../../api/comp_method.yaml
+
+name: spatrack
+label: spaTrack
+summary: "spaTrack captures fine local details of trajectory within a single tissue section of spatial transcriptomics (ST) data, as well as reconstruct cell dynamics across multiple tissue sections in a time series."
+description: | #TODO come back with detailed description of how WE are implementing it
+  To capture potential dynamic drivers, spaTrack models the fate of a cell as a function of expression profile 
+  along the time points driven by transcription factors, which facilitates the identification of key molecular 
+  regulators that govern cellular trajectories. In this task, the starting cluster is automatically determined 
+  using the functions provided by spaTrack. 
+references:
+  doi: 
+    - 10.1016/j.cels.2025.101194
+links:
+  documentation: https://spatrack.readthedocs.io/en/latest/
+  repository: https://github.com/yzf072/spaTrack
+
+
+
+info:
+  preferred_normalization: log_cp10k
+
+arguments:
+  - name: "--alpha1"
+    type: "double"
+    default: 0.5
+    description: The proportion of gene expression information.
+  - name: "--alpha2"
+    type: "double"
+    default: 0.5
+    description: The proportion of spatial location information.
+
+resources:
+  - type: python_script
+    path: script.py
+
+engines:
+  # custom image due to spatrack's python 3.9 pin
+  - type: docker
+    image: python:3.9-slim
+    setup:
+      - type: apt
+        packages:
+          - procps              # required by Nextflow
+          - git                 # pip needs it to install openproblems core from git+https
+          - build-essential     # compiler for any source builds (pysal/rtree deps)
+          - libspatialindex-dev
+          - libgeos-dev
+      - type: python
+        upgrade: true
+        github:
+          - "openproblems-bio/core#subdirectory=packages/python/openproblems"
+        packages:
+          - anndata~=0.10.0
+          - pyyaml
+          - requests
+          - jsonschema
+          - spatrack==1.0.2
+          - pot>=0.9.0
+          - scanpy>=1.9.3,<1.10 
+          - plotly>=5.15.0
+          - pygam>=0.8.1
+          - networkx>=3.0
+          - numpy==1.24.4
+          - scipy==1.10.1
+          - torch>=2.0.1
+          - pandas==1.4.3
+          - pysal==2.6.0
+
+runners:
+  - type: executable
+  - type: nextflow
+    directives:
+      label: [midtime,midmem,midcpu]

src/methods/spatrack/script.py

@@ -0,0 +1,50 @@
+import anndata as ad
+import scipy.sparse as sp
+import spaTrack as spt
+
+## VIASH START
+par = {
+  'input': 'resources_test/task_spatial_trajectory_inference/cxg_mouse_pancreas_atlas/dataset.h5ad',
+  'output': 'output.h5ad',
+  'alpha1': 0.5,
+  'alpha2': 0.5,
+}
+meta = {
+  'name': 'spatrack'
+}
+## VIASH END
+
+print('Reading input files', flush=True)
+adata = ad.read_h5ad(par['input'])
+
+adata.X = adata.layers['normalized']
+if sp.issparse(adata.X):
+    adata.X = adata.X.toarray()
+adata.obs['cluster'] = adata.obs['cell_type'].astype(str).astype('category')
+
+print('Calculate cell transition probability', flush=True)
+adata.obsp['trans'] = spt.get_ot_matrix(
+    adata,
+    data_type='spatial',
+    alpha1=par['alpha1'],
+    alpha2=par['alpha2'],
+)
+
+print('Determine starting cells', flush=True)
+adata = spt.assess_start_cluster(adata)
+start_cluster = list(adata.uns['entropy value order'].index)[0]
+start_cells = spt.set_start_cells(adata, select_way='cell_type', cell_type=start_cluster)
+
+print('Generate predictions', flush=True)
+adata.obs['pseudotime_inferred'] = spt.get_ptime(adata, start_cells)
+
+print('Write output AnnData to file', flush=True)
+output = ad.AnnData(
+    obs=adata.obs[['pseudotime_inferred']],
+    uns={
+        'dataset_id': adata.uns['dataset_id'],
+        'normalization_id': adata.uns['normalization_id'],
+        'method_id': meta['name'],
+    },
+)
+output.write_h5ad(par['output'], compression='gzip')