2727 get_interaction_pvals ,
2828 label_keys ,
2929)
30- from .ligand_receptors import LRPairs , lr_pairs
30+ from .ligand_receptors import LRPairs , genes_of , lr_pairs , split_name
3131from .neighbours import calc_distance , calc_neighbours , get_neighbourhoods
3232from .permutation import perform_spot_testing
3333from .scoring import get_lrs_scores
3434
3535
3636def _report_dropped_lrs (
3737 adata : AnnData ,
38- lrs : npt . NDArray [ np . str_ ] ,
38+ lrs : LRPairs ,
3939 n_scored : npt .NDArray [np .int64 ],
4040 min_spots : int ,
4141 min_expr : float ,
@@ -45,28 +45,31 @@ def _report_dropped_lrs(
4545 (with min_expr 0, such a pair scores in no spot; above 0 it can still score
4646 through its partner, so only dropped pairs are looked at), or else scored in
4747 min_spots spots or fewer."""
48- dropped = lrs [ n_scored <= min_spots ]
49- by_gene , unexpressed = [], set ( )
50- if dropped . size :
51- genes = np . unique ([ lr . split ( "_" ) for lr in dropped ] )
48+ dropped = lrs . filter ( n_scored <= min_spots )
49+ n_by_gene , unexpressed = 0 , np . array ( [], dtype = np . str_ )
50+ if len ( dropped ) :
51+ genes = dropped . genes ( )
5252 n_above = np .asarray ((min_expr < adata [:, genes ].X ).sum (axis = 0 )).ravel ()
53- unexpressed = set (genes [n_above == 0 ])
54- by_gene = [lr for lr in dropped if unexpressed .intersection (lr .split ("_" ))]
55- too_few = dropped .size - len (by_gene )
56- if by_gene :
57- named = ", " .join (sorted (unexpressed )[:max_named ])
53+ unexpressed = genes [n_above == 0 ] # sorted, as genes() are
54+ by_gene = np .isin (dropped .ligands (), unexpressed ) | np .isin (
55+ dropped .receptors (), unexpressed
56+ )
57+ n_by_gene = int (by_gene .sum ())
58+ too_few = len (dropped ) - n_by_gene
59+ if n_by_gene :
60+ named = ", " .join (unexpressed [:max_named ])
5861 more = (
5962 f" and { len (unexpressed ) - max_named } more"
6063 if len (unexpressed ) > max_named
6164 else ""
6265 )
6366 print (
64- f"Dropping { len ( by_gene ) } LR pairs with a gene never expressed above "
67+ f"Dropping { n_by_gene } LR pairs with a gene never expressed above "
6568 f"min_expr={ min_expr } : { named } { more } ."
6669 )
6770 if too_few :
6871 print (f"Dropping { too_few } LR pairs scored in { min_spots } or fewer spots." )
69- print (f"Altogether { lrs . size - dropped . size } valid L-R pairs" )
72+ print (f"Altogether { len ( lrs ) - len ( dropped ) } valid L-R pairs" )
7073
7174
7275def run (
@@ -238,11 +241,11 @@ def run(
238241
239242 # 1. Filter any LRs without stored expression.
240243 # Calculating the lr_scores across spots for the inputted lrs #
241- lr_scores , new_lrs = get_lrs_scores (
244+ lr_scores , scored = get_lrs_scores (
242245 adata , lrs , neighbours , het_vals , min_expr , sparse_result = True
243246 )
244247 # Fail fast.
245- if new_lrs . size == 0 :
248+ if len ( scored ) == 0 :
246249 raise ValueError (
247250 f"None of the { len (lrs )} { lrs .species } LR pairs are in adata.var_names; "
248251 "check the pairs' species matches the data, e.g. "
@@ -251,10 +254,10 @@ def run(
251254 n_scored = np .diff (lr_scores .indptr ).astype (np .int64 ) # spots scored per LR
252255 lr_bool = n_scored > min_spots
253256 if verbose :
254- _report_dropped_lrs (adata , new_lrs , n_scored , min_spots , min_expr )
255- new_lrs = new_lrs [ lr_bool ]
257+ _report_dropped_lrs (adata , scored , n_scored , min_spots , min_expr )
258+ tested = scored . filter ( lr_bool )
256259 lr_scores = lr_scores [:, lr_bool ]
257- if new_lrs . size == 0 :
260+ if len ( tested ) == 0 :
258261 raise ValueError (
259262 f"No LR pairs had scores in more than { min_spots } spots; "
260263 "try lowering min_spots or min_expr."
@@ -263,7 +266,7 @@ def run(
263266 tests = perform_spot_testing (
264267 adata ,
265268 lr_scores ,
266- new_lrs ,
269+ tested ,
267270 n_pairs ,
268271 neighbours ,
269272 het_vals ,
@@ -274,10 +277,10 @@ def run(
274277 random_state = random_state ,
275278 )
276279 lr_results = AnnData (
277- obs = pd .DataFrame (index = adata .obs_names ), var = pd .DataFrame (index = new_lrs )
280+ obs = pd .DataFrame (index = adata .obs_names ), var = pd .DataFrame (index = tested . names () )
278281 )
279282
280- lr_results .var ["source" ] = lrs . source_of ( new_lrs ). astype ( str )
283+ lr_results .var ["source" ] = tested . to_frame ()[ "source" ]. to_numpy ( dtype = str )
281284 lr_results .uns ["lr_species" ] = lrs .species
282285 lr_results .uns ["lrfeatures" ] = tests .lr_features
283286
@@ -398,7 +401,7 @@ def run_lr_go(
398401 lrs = lr_results .var_names .to_numpy ().astype (str )
399402 n_sig = lr_results .var ["n_spots_sig" ].to_numpy ().astype (int )
400403 top_lrs = lrs [n_sig > min_sig_spots ][0 :n_top ]
401- top_genes = np . unique ([ lr . split ( "_" ) for lr in top_lrs ] )
404+ top_genes = genes_of ( top_lrs )
402405
403406 # Determining the background genes if not inputted
404407 if bg_genes is None :
@@ -571,7 +574,7 @@ def run_cci(
571574 col = "lr_sig_scores" if sig_spots else "lr_scores"
572575 best_lr_indices = np .flatnonzero (lr_summary [count_col ].to_numpy () > min_spots )
573576 best_lrs = lr_summary .index .values [best_lr_indices ]
574- lr_genes = np . unique ([ lr . split ( "_" ) for lr in best_lrs ] )
577+ lr_genes = genes_of ( best_lrs )
575578 if len (lr_genes ) == 0 :
576579 raise ValueError (
577580 "No LR pairs returned with current filtering params; \n "
@@ -602,7 +605,7 @@ def expressed(gene: str) -> npt.NDArray[np.bool_]:
602605 disable = not verbose ,
603606 ) as pbar :
604607 for i , best_lr in enumerate (best_lrs ):
605- ligand , receptor = best_lr . split ( "_" )
608+ ligand , receptor = split_name ( best_lr )
606609
607610 L_bool = expressed (ligand )
608611 R_bool = expressed (receptor )
0 commit comments