From 3009725f618aabb611ec593cc0dea10e27a743d1 Mon Sep 17 00:00:00 2001
From: "C. Titus Brown" <titus@idyll.org>
Date: Wed, 19 Jun 2024 06:39:56 -0700
Subject: [PATCH] MRG: adjust how ANI is calculated in the revindex code.
 (#3218)

Calculate ANI of matches against original query with `f_orig_query` and
`f_match_orig`, instead of against `f_unique_to_query` and `f_match`.

This fixes the ANI differences between `sourmash gather` and RocksDB
branchwater gather for the columns `query_containment_ani`,
`match_containment_ani`, `max_containment_ani`, and
`average_containment_ani`.

Refs:
* Used by
https://github.com/sourmash-bio/sourmash_plugin_branchwater/pull/361
* Fixes RocksDB-based calculations for
https://github.com/sourmash-bio/sourmash_plugin_branchwater/issues/331
---
 src/core/Cargo.toml                | 2 +-
 src/core/src/index/mod.rs          | 4 ++--
 src/core/src/index/revindex/mod.rs | 3 +++
 3 files changed, 6 insertions(+), 3 deletions(-)

diff --git a/src/core/Cargo.toml b/src/core/Cargo.toml
index 71249863a1..c89cb0d28d 100644
--- a/src/core/Cargo.toml
+++ b/src/core/Cargo.toml
@@ -44,7 +44,7 @@ md5 = "0.7.0"
 memmap2 = "0.9.4"
 murmurhash3 = "0.0.5"
 needletail = { version = "0.5.1", default-features = false }
-niffler = { version = "2.6.0", default-features = false, features = [ "gz" ] }
+niffler = { version = "2.4.0", default-features = false, features = [ "gz" ] }
 nohash-hasher = "0.2.0"
 num-iter = "0.1.45"
 once_cell = "1.18.0"
diff --git a/src/core/src/index/mod.rs b/src/core/src/index/mod.rs
index 3ce2652808..4816a56cc8 100644
--- a/src/core/src/index/mod.rs
+++ b/src/core/src/index/mod.rs
@@ -244,8 +244,8 @@ pub fn calculate_gather_stats(
 
     // // get ANI values
     let ksize = match_mh.ksize() as f64;
-    let query_containment_ani = ani_from_containment(f_unique_to_query, ksize);
-    let match_containment_ani = ani_from_containment(f_match, ksize);
+    let query_containment_ani = ani_from_containment(f_orig_query, ksize);
+    let match_containment_ani = ani_from_containment(f_match_orig, ksize);
     let mut query_containment_ani_ci_low = None;
     let mut query_containment_ani_ci_high = None;
     let mut match_containment_ani_ci_low = None;
diff --git a/src/core/src/index/revindex/mod.rs b/src/core/src/index/revindex/mod.rs
index fc03896385..8e2b35f716 100644
--- a/src/core/src/index/revindex/mod.rs
+++ b/src/core/src/index/revindex/mod.rs
@@ -814,6 +814,7 @@ mod test {
         assert_eq!(round5(match_.f_unique_to_query()), round5(0.13096862));
         assert_eq!(match_.unique_intersect_bp, 1920000);
         assert_eq!(match_.remaining_bp, 12740000);
+        assert_eq!(round5(match_.query_containment_ani()), round5(0.90773763));
 
         let match_ = &matches[1];
         let names: Vec<&str> = match_.name().split(' ').take(1).collect();
@@ -822,6 +823,7 @@ mod test {
         assert_eq!(round5(match_.f_unique_to_query()), round5(0.115279));
         assert_eq!(match_.unique_intersect_bp, 1690000);
         assert_eq!(match_.remaining_bp, 11050000);
+        assert_eq!(round5(match_.query_containment_ani()), round5(0.9068280));
 
         let match_ = &matches[2];
         dbg!(match_);
@@ -831,6 +833,7 @@ mod test {
         assert_eq!(round5(match_.f_unique_to_query()), round5(0.0627557));
         assert_eq!(match_.unique_intersect_bp, 920000);
         assert_eq!(match_.remaining_bp, 10130000);
+        assert_eq!(round5(match_.query_containment_ani()), round5(0.90728512));
 
         Ok(())
     }