From 29da1240b7b81d86a904fdc6c740ecb658fd4e61 Mon Sep 17 00:00:00 2001
From: William Fondrie <fondriew@gmail.com>
Date: Wed, 26 Jun 2024 15:07:51 -0700
Subject: [PATCH] Added simple system test

---
 petasus/scripts/sage2lib.py         |  6 +++++-
 tests/data/results.sage.csv         |  2 ++
 tests/system_tests/test_sage2lib.py | 29 +++++++++++++++++++++++++++++
 3 files changed, 36 insertions(+), 1 deletion(-)
 create mode 100644 tests/data/results.sage.csv
 create mode 100644 tests/system_tests/test_sage2lib.py

diff --git a/petasus/scripts/sage2lib.py b/petasus/scripts/sage2lib.py
index 2eb6da6..816e167 100644
--- a/petasus/scripts/sage2lib.py
+++ b/petasus/scripts/sage2lib.py
@@ -24,7 +24,11 @@ def read_peptides(peptides, qvalue):
     polars.DataFrame
         The parsed and filtered peptides.
     """
-    peptide_df = pl.read_csv(peptides, separator="\t")
+    try:
+        peptide_df = pl.read_parquet(peptides)
+    except pl.exceptions.ComputeError:
+        peptide_df = pl.read_csv(peptides, separator="\t")
+
     peptide_df.columns = [c.lower() for c in peptide_df.columns]
     if "mokapot q-value" in peptide_df.columns:
         qval_col = "mokapot q-value"
diff --git a/tests/data/results.sage.csv b/tests/data/results.sage.csv
new file mode 100644
index 0000000..1c00cc9
--- /dev/null
+++ b/tests/data/results.sage.csv
@@ -0,0 +1,2 @@
+filename	scannr	peptide	stripped_peptide	proteins	num_proteins	rank	is_decoy	expmass	calcmass	charge	peptide_len	missed_cleavages	isotope_error	precursor_ppm	fragment_ppm	hyperscore	delta_next	delta_best	rt	aligned_rt	predicted_rt	delta_rt_model	matched_peaks	longest_b	longest_y	longest_y_pct	matched_intensity_pct	scored_candidates	poisson	sage_discriminant_score	posterior_error	spectrum_q	peptide_q	protein_q
+LQSRPAAPPAPGPGQLTLR.mzML	controllerType=0 controllerNumber=1 scan=30069	LQSRPAAPPAPGPGQLTLR	LQSRPAAPPAPGPGQLTLR	sp|Q99536|VAT1_HUMAN	1	1	false	1926.0815	1926.08	3	19	0	0.0	0.8239083	0.503857	72.265915	72.265915	0.0	108.2854	0.993444	0.0	0.993444	22	9	12	0.6315789	64.770966	1	-1.9562812	1.2944585	1.0	1.0	1.0	1.0
diff --git a/tests/system_tests/test_sage2lib.py b/tests/system_tests/test_sage2lib.py
new file mode 100644
index 0000000..f6ed9c1
--- /dev/null
+++ b/tests/system_tests/test_sage2lib.py
@@ -0,0 +1,29 @@
+"""Test building a dlib"""
+import sqlite3
+import subprocess
+
+import pytest
+
+
+@pytest.mark.parametrize("ftype", ["parquet", "csv"])
+def test_simpole_speclib_generation(data_path, tmp_path, ftype):
+    """Test spectral library generation"""
+    psms = data_path / f"results.sage.{ftype}"
+    mzml = data_path / "LQSRPAAPPAPGPGQLTLR.mzML"
+    outfile = tmp_path / "results.sage.dlib"
+    cmd = [
+        "petasus",
+        "sage2lib",
+        "--qvalue",
+        "1",
+        f"{psms}",
+        f"{mzml}",
+    ]
+
+    subprocess.run(cmd, check=True)
+
+    con = sqlite3.Connection(str(outfile))
+    cur = con.cursor()
+    res = cur.execute("select PeptideModSeq from entries;")
+    res = [x[0] for x in res.fetchall()]
+    assert len(res)