Merge branch 'feat/sparse-dist' into 'master'

Scale up TCR dist

Closes #38 and #41

See merge request icbi-lab/pipelines/singlecell_tcr!22

Former-commit-id: 0b60de2d242c8c5297fb676e162e1d06e6eadfe2

docs/api.rst

@@ -42,6 +42,8 @@ Preprocessing: `pp`
    :toctree: .
 
    merge_with_tcr
+   tcr_dist
+   tcr_neighbors
 
 
 Tools: `tl`
@@ -54,7 +56,6 @@ Tools: `tl`
 
    define_clonotypes
    clonotype_network
-   tcr_dist
    chain_pairing
    clip_and_count
    alpha_diversity

docs/tutorials/tutorial_3k_tcr.md

@@ -53,10 +53,6 @@ It just has additional TCR-related columns in `obs`:
 adata.obs
 ```
 
-```python
-
-```
-
 ## Preprocess Transcriptomics data
 
 Transcriptomics data needs to be filtered and preprocessed as with any other single-cell dataset.
@@ -128,7 +124,11 @@ All connected nodes form a clonotype.
 * With the `chains` parameter, we can specify if we want to consider only the most abundant TRA and TRB sequences (`primary_only`), or all four CDR3 sequences, if available (`all`). 
 
 ```python
-st.tl.define_clonotypes(adata, strategy="all", chains="primary_only", cutoff=0)
+st.pp.tcr_neighbors(adata, strategy="all", chains="primary_only", cutoff=0)
+```
+
+```python
+st.tl.define_clonotypes(adata)
 ```
 
 Let's visualize the resulting graph. 
@@ -148,7 +148,8 @@ Now, we allow a TCR-distance of 20. That's the equivalent of 4 `R`s mutating int
 Also we now use `chains='all'`
 
 ```python
-st.tl.define_clonotypes(adata, strategy="all", chains="all", cutoff=20)
+st.pp.tcr_neighbors(adata, strategy="all", chains="all", cutoff=20)
+st.tl.define_clonotypes(adata)
 ```
 
 ```python

sctcrpy/_plotting/_tcr_dist.py

@@ -2,13 +2,15 @@
 from anndata import AnnData
 import igraph as ig
 import numpy as np
-from .._tools._tcr_dist import get_igraph_from_adjacency
+from .._util import get_igraph_from_adjacency
 
 
-def clonotype_network(adata, color, *, key="clonotype", obsm_key="X_clonotype_network"):
+def clonotype_network(
+    adata, color, *, neighbors_key="neighbors", obsm_key="X_clonotype_network"
+):
     """Plot the clonotype network"""
     idx = np.where(~np.any(np.isnan(adata.obsm[obsm_key]), axis=1))[0]
-    adj = adata.uns["sctcrpy"][key + "_connectivities"][idx, :][:, idx]
+    adj = adata.uns["sctcrpy"][neighbors_key]["connectivities"][idx, :][:, idx]
     g = get_igraph_from_adjacency(adj)
     layout = ig.Layout(adata.obsm[obsm_key][idx, :].tolist())
     # return g, layout

sctcrpy/_preprocessing/__init__.py

@@ -0,0 +1,2 @@
+from ._merge_adata import merge_with_tcr
+from ._tcr_dist import tcr_dist, tcr_neighbors