From 9a0e133f4e86eb9d335050227d0e9f6d6fa6f718 Mon Sep 17 00:00:00 2001
From: Jerome Kelleher <jk@well.ox.ac.uk>
Date: Fri, 20 Sep 2024 17:26:57 +0100
Subject: [PATCH] Move recombinant processing offline

Because it slows down inference a lot. We have only a few recombinants,
so we don't effectively parallelise, and we do it three times. Also,
we're doing some expensive matching on things which will mostly not
get into the final ARG.
---
 sc2ts/inference.py      | 13 +-------
 tests/test_inference.py | 69 ++---------------------------------------
 2 files changed, 3 insertions(+), 79 deletions(-)

diff --git a/sc2ts/inference.py b/sc2ts/inference.py
index f6b7689..73fc74b 100644
--- a/sc2ts/inference.py
+++ b/sc2ts/inference.py
@@ -461,21 +461,10 @@ def match_samples(
         cost = hmm_match.get_hmm_cost(num_mismatches)
         # print(f"Final HMM pass:{sample.strain} hmm_cost={cost} {sample.summary()}")
         logger.debug(f"Final HMM pass hmm_cost={cost} {sample.summary()}")
-        if sample.is_recombinant:
-            recombinants.append(sample)
-
-    if len(recombinants) > 0:
-        match_recombinants(
-            recombinants,
-            base_ts,
-            num_mismatches=num_mismatches,
-            show_progress=show_progress,
-            num_threads=num_threads,
-        )
-
     return samples
 
 
+
 def check_base_ts(ts):
     md = ts.metadata
     assert "sc2ts" in md
diff --git a/tests/test_inference.py b/tests/test_inference.py
index 12af229..6559356 100644
--- a/tests/test_inference.py
+++ b/tests/test_inference.py
@@ -630,37 +630,7 @@ def test_recombinant_example_1(self, tmp_path, fx_ts_map, fx_alignment_store):
                 {"left": 15324, "parent": 52, "right": 29904},
             ],
         }
-        assert smd["hmm_reruns"] == {
-            "forward": {
-                "mutations": [],
-                "path": [
-                    {"left": 0, "parent": 36, "right": 15324},
-                    {"left": 15324, "parent": 52, "right": 29904},
-                ],
-            },
-            "no_recombination": {
-                "mutations": [
-                    {
-                        "derived_state": "T",
-                        "inherited_state": "C",
-                        "site_position": 15324,
-                    },
-                    {
-                        "derived_state": "T",
-                        "inherited_state": "C",
-                        "site_position": 29303,
-                    },
-                ],
-                "path": [{"left": 0, "parent": 36, "right": 29904}],
-            },
-            "reverse": {
-                "mutations": [],
-                "path": [
-                    {"left": 0, "parent": 36, "right": 3788},
-                    {"left": 3788, "parent": 52, "right": 29904},
-                ],
-            },
-        }
+        assert smd["hmm_reruns"] == {}
 
         recomb_node = ts.node(ts.num_nodes - 1)
         assert recomb_node.flags == sc2ts.NODE_IS_RECOMBINANT
@@ -755,42 +725,7 @@ def test_recombinant_example_2(self, tmp_path, fx_ts_map, fx_alignment_store):
             ],
         }
 
-        assert smd["hmm_reruns"] == {
-            "forward": {
-                "mutations": [],
-                "path": [
-                    {"left": 0, "parent": 62, "right": 29800},
-                    {"left": 29800, "parent": 63, "right": 29904},
-                ],
-            },
-            "no_recombination": {
-                "mutations": [
-                    {
-                        "derived_state": "A",
-                        "inherited_state": "G",
-                        "site_position": 29800,
-                    },
-                    {
-                        "derived_state": "A",
-                        "inherited_state": "C",
-                        "site_position": 29801,
-                    },
-                    {
-                        "derived_state": "A",
-                        "inherited_state": "C",
-                        "site_position": 29802,
-                    },
-                ],
-                "path": [{"left": 0, "parent": 62, "right": 29904}],
-            },
-            "reverse": {
-                "mutations": [],
-                "path": [
-                    {"left": 0, "parent": 62, "right": 113},
-                    {"left": 113, "parent": 63, "right": 29904},
-                ],
-            },
-        }
+        assert smd["hmm_reruns"] == { }
 
     def test_all_As(self, tmp_path, fx_ts_map, fx_alignment_store):
         # Same as the recombinant_example_1() function above