Fix pandas IndexError in export matrix summarization (pandas ≥2.1)

[Copilot]

pyprophet export matrix crashes with IndexError: DataFrame indexer is not allowed for .iloc on pandas ≥2.1, affecting --level=peptide, --level=protein, and --level=precursor exports.

Root Cause

groupby().apply(lambda x: x["m_score"].idxmin()) returns a Series of index labels, not integer positions. Passing this to .iloc[] was silently accepted in older pandas but is explicitly rejected in ≥2.1.

Changes

• pyprophet/io/_base.py — two instances fixed (_summarize_precursor_level and _summarize_peptide_level): replace .iloc[groupby().apply(lambda x: x[...].idxmin())] with the correct and more efficient .loc[groupby()[...].idxmin()]

# Before (broken in pandas ≥2.1):
data = data.iloc[
    data.groupby(["run_id", "transition_group_id"]).apply(
        lambda x: x["m_score"].idxmin()
    )
]

# After:
idx = data.groupby(["run_id", "transition_group_id"])["m_score"].idxmin()
data = data.loc[idx]

Original prompt

---

This section details on the original issue you should resolve

<issue_title>export matrix: pandas IndexError in _summarize_peptide_level (v3.0.7)</issue_title>
<issue_description>## Description

pyprophet export matrix crashes with a pandas IndexError in _summarize_peptide_level (file pyprophet/io/_base.py, line ~804). This affects both --level=peptide and --level=protein exports.

Environment

• pyprophet 3.0.7
• pandas 2.1.4 (also reproducible with 2.3.3)
• Python 3.10.19
• Linux

Error

File "pyprophet/io/_base.py", line 804, in _summarize_peptide_level
    data = data.iloc[
        data.groupby(["run_id", "transition_group_id"]).apply(
            lambda x: x["m_score"].idxmin()
        )
    ]
IndexError: DataFrame indexer is not allowed for .iloc
Consider using .loc for automatic alignment.

Root cause

groupby().apply() returns a Series object. In recent pandas versions (>=2.1), passing a Series to .iloc[] is no longer allowed. The fix is to use .loc[] with idxmin() directly:

# Before (broken):
data = data.iloc[
    data.groupby(["run_id", "transition_group_id"]).apply(
        lambda x: x["m_score"].idxmin()
    )
]

# After (fixed):
idx = data.groupby(["run_id", "transition_group_id"])["m_score"].idxmin()
data = data.loc[idx]

Note: there is a second instance of the same .iloc pattern in the same function (for _summarize_protein_level) that should be fixed as well.

Steps to reproduce

1. Run pyprophet score on an .osw file
2. Run pyprophet export matrix --level=peptide

The crash occurs in _summarize_peptide_level regardless of the data content.</issue_description>

<agent_instructions>Can you look into this issue.</agent_instructions>

Comments on the Issue (you are @copilot in this section)

• Fixes export matrix: pandas IndexError in _summarize_peptide_level (v3.0.7) #193

---

💬 We'd love your input! Share your thoughts on Copilot coding agent in our 2 minute survey.

[Copilot]

Pull request overview

This PR fixes a pandas ≥2.1 IndexError crash that occurred when running pyprophet export matrix at any summarization level. The root cause was using .iloc[] with a Series of index labels (returned by groupby().apply(lambda x: x[col].idxmin())), which was silently accepted in older pandas but explicitly rejected in ≥2.1. The fix replaces this with the equivalent and more efficient .loc[groupby()[col].idxmin()].

Changes:

• Replace .iloc[groupby().apply(lambda x: x["m_score"].idxmin())] with .loc[groupby()["m_score"].idxmin()] in _summarize_precursor_level
• Apply the same fix in _summarize_peptide_level

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

The same IndexError: DataFrame indexer is not allowed for .iloc bug that this PR fixes in pyprophet/io/_base.py also exists in pyprophet/export/export_compound.py (lines 179-183) and pyprophet/glyco/export.py (lines 501-505). Both files still use the broken pattern:

data = data.iloc[
    data.groupby(["run_id", "transition_group_id"]).apply(
        lambda x: x["m_score"].idxmin()
    )
]

These should be updated to use the same fix applied here:

idx = data.groupby(["run_id", "transition_group_id"])["m_score"].idxmin()
data = data.loc[idx]

These will crash with pandas ≥2.1 exactly as described in the issue.

[Copilot]

There are no tests for export matrix (the export_quant_matrix path via _summarize_precursor_level, _summarize_peptide_level, _summarize_protein_level, and _summarize_gene_level). The existing test file tests/test_pyprophet_export.py covers many other export paths. Adding at least one integration test that exercises pyprophet export matrix --level=precursor (and ideally --level=peptide) would prevent regressions like this from going undetected.