From 8d138b33170034911c7fc174525e7565a1d54d09 Mon Sep 17 00:00:00 2001
From: Tessa Pierce Ward <bluegenes@users.noreply.github.com>
Date: Mon, 19 Feb 2024 08:17:02 -0800
Subject: [PATCH] k/3 for manifest to reflect input ksize (#224)

---
 src/python/tests/test_sketch.py | 44 +++++++++++++++++++++++++++++++++
 src/utils.rs                    |  7 +++++-
 2 files changed, 50 insertions(+), 1 deletion(-)

diff --git a/src/python/tests/test_sketch.py b/src/python/tests/test_sketch.py
index d6bbc5fa..fdbad0b9 100644
--- a/src/python/tests/test_sketch.py
+++ b/src/python/tests/test_sketch.py
@@ -405,3 +405,47 @@ def test_zip_manifest(runtmp, capfd):
         assert sig.minhash.ksize == 31
         assert sig.minhash.moltype == 'DNA'
         assert sig.minhash.scaled == 1
+
+
+def test_protein_zip_manifest(runtmp, capfd):
+    # test basic manifest-generating functionality.
+    fa_csv = runtmp.output('db-fa.csv')
+
+    fa1 = get_test_data('short.fa')
+    fa2 = get_test_data('short-protein.fa')
+
+    make_file_csv(fa_csv, [fa1], [fa2])
+    output = runtmp.output('db.zip')
+
+    runtmp.sourmash('scripts', 'manysketch', fa_csv, '-o', output,
+                    '--param-str', "protein,k=10,scaled=1")
+
+    loader = sourmash.load_file_as_index(output)
+
+    rows = []
+    siglist = []
+    # make manifest via sourmash python code
+    for (sig, loc) in loader._signatures_with_internal():
+        row = index.CollectionManifest.make_manifest_row(sig, loc)
+        rows.append(row)
+        siglist.append(sig)
+
+    manifest = index.CollectionManifest(rows)
+
+    assert len(manifest) == len(rows)
+    assert len(manifest) == 1
+
+    md5_list = [ row['md5'] for row in manifest.rows ]
+    assert 'eb4467d11e0ecd2dbde4193bfc255310' in md5_list
+    ksize_list = [ row['ksize'] for row in manifest.rows ]
+    assert 10 in ksize_list # manifest ksizes are human-readable (k, not k*3)
+    scaled_list = [ row['scaled'] for row in manifest.rows ]
+    assert 1 in scaled_list
+    moltype_list = [ row['moltype'] for row in manifest.rows ]
+    assert "protein" in moltype_list
+
+    for sig in siglist:
+        assert sig in manifest
+        assert sig.minhash.ksize == 10 # minhash stores k*3, but does the conversion back for us
+        assert sig.minhash.moltype == 'protein'
+        assert sig.minhash.scaled == 1
diff --git a/src/utils.rs b/src/utils.rs
index 7342af97..db2c84e4 100644
--- a/src/utils.rs
+++ b/src/utils.rs
@@ -787,11 +787,16 @@ pub fn make_manifest_row(
         "protein".to_string()
     };
     let sketch = &sig.sketches()[0];
+    let ksize: u32 = if is_dna {
+        sketch.ksize() as u32
+    } else {
+        sketch.ksize() as u32 / 3
+    };
     ManifestRow {
         internal_location: internal_location.to_string(),
         md5: sig.md5sum(),
         md5short: sig.md5sum()[0..8].to_string(),
-        ksize: sketch.ksize() as u32,
+        ksize: ksize,
         moltype,
         num,
         scaled,