Merge pull request #618 from bigbio/dev

Include the metadata and version of DIANN in the reports from log files

Author: ypriverol

.github/workflows/python-app.yml

@@ -98,7 +98,7 @@ jobs:
         run: |
           wget -nv https://ftp.pride.ebi.ac.uk/pub/databases/pride/resources/proteomes/pmultiqc/dia/dia.zip
           unzip -d ./dia dia.zip
-          multiqc --quantms-plugin ./dia --config  ./dia/multiqc_config.yml -o ./results_dia
+          multiqc --quantms-plugin ./dia --config ./dia/multiqc_config.yml -o ./results_dia
       - uses: actions/upload-artifact@v4
         if: always()
         name: Upload results

README.md

@@ -50,6 +50,7 @@ pmultiqc supports the following data sources:
 
 3. **[DIA-NN](https://aptila.bio)** result files:
    - `report.tsv` or `report.parquet`: DIA-NN main report
+   - `report.log.txt` or `diannsummary.log`: DIA-NN log
    - `*sdrf.tsv`: SDRF-Proteomics (optional)
    - `*ms_info.parquet`: mzML statistics after RAW-to-mzML conversion (using **[quantms-utils](https://github.com/bigbio/quantms-utils)**) (optional)
 

docs/README.md

@@ -70,6 +70,7 @@ pmultiqc supports the following data sources:
 
 ### 3. **[DIA-NN](https://aptila.bio)** result files:
    - `report.tsv` or `report.parquet`: DIA-NN main report
+   - `report.log.txt` or `diannsummary.log`: DIA-NN log
    - `*sdrf.tsv`: SDRF-Proteomics (optional)
    - `*ms_info.parquet`: mzML statistics after RAW-to-mzML conversion (using **[quantms-utils](https://github.com/bigbio/quantms-utils)**) (optional)
 

pmultiqc/main.py

@@ -135,6 +135,12 @@ def pmultiqc_plugin_execution_start():
             config.sp, {"pmultiqc/diann_report_parquet": {"fn": "report.parquet", "num_lines": 0}}
         )
 
+    if "pmultiqc/diann_log_txt" not in config.sp:
+        config.update_dict(config.sp, {"pmultiqc/diann_log_txt": {"fn": "report.log.txt", "num_lines": 0}})
+
+    if "pmultiqc/diann_log" not in config.sp:
+        config.update_dict(config.sp, {"pmultiqc/diann_log": {"fn": "diannsummary.log", "num_lines": 0}})
+
     if "pmultiqc/maxquant_result" not in config.sp:
         config.update_dict(
             config.sp, {"pmultiqc/maxquant_result": {"fn": "*.txt", "num_lines": 0}}

pmultiqc/modules/common/dia_utils.py

@@ -5,6 +5,7 @@
 from collections import OrderedDict
 from sdrf_pipelines.openms.openms import UnimodDatabase
 from multiqc.plots import table
+import os
 
 from pmultiqc.modules.common.histogram import Histogram
 from pmultiqc.modules.common.stats import (
@@ -984,3 +985,77 @@ def dia_sample_level_modifications(df, sdrf_file_df):
         mod_plot[f"Sample {str(sample)}"] = mod_plot_dict
 
     return mod_plot
+
+
+def parse_diann_version(log_file_path):
+    """Parse the DIA-NN version from a DIA-NN log file (report.log.txt).
+
+    The DIA-NN log file typically starts with a line like:
+        DIA-NN 1.8.1 (Data-Independent Acquisition by Neural Networks)
+
+    Args:
+        log_file_path: Path to the DIA-NN log file.
+
+    Returns:
+        Version string (e.g. "1.8.1") or None if not found.
+    """
+    version_pattern = re.compile(r"^DIA-NN\s+(\d+(?:\.\d+)*)", re.IGNORECASE)
+    try:
+        with open(log_file_path, "r", encoding="utf-8", errors="replace") as f:
+            for i, line in enumerate(f):
+                match = version_pattern.match(line.strip())
+                if match:
+                    return match.group(1)
+                # Version line is expected near the top; stop after 20 lines
+                if i >= 20:
+                    break
+    except Exception as e:
+        log.warning(f"Could not parse DIA-NN version from {log_file_path}: {e}")
+    return None
+
+
+def draw_diann_metadata_table(sub_section, diann_version):
+    """Draw a metadata table showing DIA-NN software version.
+
+    Args:
+        sub_section: The sub-section list to add the table to.
+        diann_version: DIA-NN version string to display.
+    """
+    if not diann_version:
+        return
+
+    table_data = {
+        1: {
+            "parameter": "DIA-NN Version",
+            "value": diann_version,
+        }
+    }
+
+    draw_config = {
+        "id": "diann_metadata",
+        "title": "DIA-NN Metadata",
+        "save_file": False,
+        "sort_rows": False,
+        "only_defined_headers": True,
+        "col1_header": "No.",
+        "no_violin": True,
+        "save_data_file": False,
+    }
+
+    headers = {
+        "parameter": {"title": "Parameter"},
+        "value": {"title": "Value"},
+    }
+
+    table_html = table.plot(data=table_data, headers=headers, pconfig=draw_config)
+
+    add_sub_section(
+        sub_section=sub_section,
+        plot=table_html,
+        order=0,
+        description="This table presents the DIA-NN software version used for the analysis.",
+        helptext="""
+            DIA-NN metadata, extracted from the DIA-NN log file (report.log.txt), shows the
+            version of DIA-NN used for data-independent acquisition analysis.
+            """,
+    )

pmultiqc/modules/diann/diann.py

@@ -13,7 +13,7 @@
     parse_mzml,
     aggregate_general_stats
 )
-from pmultiqc.modules.common.dia_utils import parse_diann_report
+from pmultiqc.modules.common.dia_utils import parse_diann_report, parse_diann_version, draw_diann_metadata_table
 from pmultiqc.modules.common.plots.general import draw_exp_design
 from pmultiqc.modules.common.plots.id import (
     draw_summary_protein_ident_table,
@@ -98,6 +98,20 @@ def get_data(self):
             log.error("DIANN report not found. Please check your data!")
             return False
 
+        # DIA-NN log file for version extraction
+        self.diann_version = None
+        for key in ["pmultiqc/diann_log_txt", "pmultiqc/diann_log"]:
+            for f in self.find_log_files(key, filecontents=False):
+                log_path = os.path.join(f["root"], f["fn"])
+                self.diann_version = parse_diann_version(log_path)
+
+                if self.diann_version:
+                    log.info(f"DIA-NN version detected: {self.diann_version}")
+                    break
+
+            if self.diann_version:
+                break
+
         (
             self.mzml_table,
             self.mzml_peaks_ms2_plot,
@@ -133,6 +147,10 @@ def draw_plots(self):
         self.cal_num_table_data = {}
         self.quantms_modified = {}
 
+        # Draw DIA-NN metadata table (version info) in the experiment section
+        if self.diann_version:
+            draw_diann_metadata_table(self.sub_sections["experiment"], self.diann_version)
+
         general_stats_data = aggregate_general_stats(
             ms1_general_stats=self.ms1_general_stats,
             current_sum_by_run=self.current_sum_by_run,

pmultiqc/modules/quantms/quantms.py

@@ -29,7 +29,7 @@
 from . import sparklines
 
 from pmultiqc.modules.base import BasePMultiqcModule
-from pmultiqc.modules.common.dia_utils import parse_diann_report
+from pmultiqc.modules.common.dia_utils import parse_diann_report, parse_diann_version, draw_diann_metadata_table
 from pmultiqc.modules.common.common_utils import (
     parse_sdrf,
     get_ms_path,
@@ -255,6 +255,15 @@ def get_data(self):
             self.diann_report_path = diann_report_path
             self.enable_dia = True
 
+        # DIA-NN log file for version extraction
+        self.diann_version = None
+        for f in self.find_log_files("pmultiqc/diann_log", filecontents=False):
+            log_path = os.path.join(f["root"], f["fn"])
+            self.diann_version = parse_diann_version(log_path)
+            if self.diann_version:
+                log.info(f"DIA-NN version detected: {self.diann_version}")
+                break
+
         if not self.enable_dia:
             for f in self.find_log_files("pmultiqc/mztab", filecontents=False):
                 self.out_mztab_path = os.path.join(f["root"], f["fn"])
@@ -316,6 +325,10 @@ def draw_plots(self):
         # quantms: DIA
         if self.enable_dia:
 
+            # Draw DIA-NN metadata table (version info) in the experiment section
+            if self.diann_version:
+                draw_diann_metadata_table(self.sub_sections["experiment"], self.diann_version)
+
             (
                 self.total_protein_quantified,
                 self.total_peptide_count,

tests/resources/diann/report.log.txt

@@ -0,0 +1,7 @@
+DIA-NN 1.8.1 (Data-Independent Acquisition by Neural Networks)
+Compiled on Jun 23 2022 at 15:23:37
+
+Thread number set to 1
+Library-free mode is not active
+Mass accuracy settings detected
+Loading FASTA /data/uniprot_human.fasta

tests/test_diann.py

@@ -0,0 +1,44 @@
+"""Tests for DIA-NN module functionality."""
+
+import os
+from pathlib import Path
+
+import pytest
+
+from pmultiqc.modules.common.dia_utils import parse_diann_version
+
+TEST_DATA_DIR = Path(os.path.dirname(__file__)) / "resources" / "diann"
+
+
+class TestDiann:
+    """Tests for DIA-NN module functions."""
+
+    def test_parse_diann_version_from_log(self):
+        """Test parsing DIA-NN version from log file."""
+        log_file = TEST_DATA_DIR / "report.log.txt"
+        assert log_file.exists(), f"Test log file {log_file} does not exist"
+
+        version = parse_diann_version(str(log_file))
+        assert version is not None, "DIA-NN version should be parsed from log file"
+        assert version == "1.8.1", f"Expected version '1.8.1', got '{version}'"
+
+    def test_parse_diann_version_nonexistent_file(self):
+        """Test that parse_diann_version returns None for nonexistent file."""
+        version = parse_diann_version("/nonexistent/path/report.log.txt")
+        assert version is None, "Should return None for nonexistent file"
+
+    def test_parse_diann_version_no_version_line(self, tmp_path):
+        """Test that parse_diann_version returns None when no version line is found."""
+        log_file = tmp_path / "report.log.txt"
+        log_file.write_text("This log file has no version line.\nAnother line.\n")
+
+        version = parse_diann_version(str(log_file))
+        assert version is None, "Should return None when no version line is present"
+
+    def test_parse_diann_version_different_formats(self, tmp_path):
+        """Test parsing DIA-NN version with different version string formats."""
+        log_file = tmp_path / "report.log.txt"
+        log_file.write_text("DIA-NN 2.0 (Data-Independent Acquisition by Neural Networks)\n")
+
+        version = parse_diann_version(str(log_file))
+        assert version == "2.0", f"Expected version '2.0', got '{version}'"