masster 0.6.1__py3-none-any.whl → 0.6.2__py3-none-any.whl

masster/sample/save.py

@@ -808,7 +808,7 @@ def export_dda_stats(self, filename="stats.csv"):
     self.logger.success(f"DDA statistics exported to {filename}")
 
 
-def export_xlsx(self, filename="features.xlsx"):
+def export_excel(self, filename="features.xlsx"):
     """
     Export the features DataFrame to an Excel file.
 
@@ -857,6 +857,143 @@ def export_xlsx(self, filename="features.xlsx"):
     self.logger.debug(f"Exported {len(clean_df)} features with {len(exportable_columns)} columns")
 
 
+def export_slaw(self, filename="features_slaw.csv"):
+    """
+    Export the features DataFrame to a SLAW-formatted CSV file.
+
+    This method exports the features to a tab-separated CSV format compatible with SLAW,
+    including feature metadata and intensity quantification. The file contains comprehensive
+    feature information including m/z, RT, annotations, isotopic patterns, MS2 data, and
+    intensity values.
+
+    Parameters:
+        filename (str): The path to the output CSV file. Defaults to 'features_slaw.csv'.
+
+    Side Effects:
+        Writes the exported data to the specified CSV file and logs the export operation.
+    """
+    if self.features_df is None:
+        self.logger.warning("No features found. Cannot export to SLAW format.")
+        return
+
+    filename = os.path.abspath(filename)
+
+    # Get base filename for quant column
+    if self.file_path is not None:
+        base_name = os.path.splitext(os.path.basename(self.file_path))[0]
+    else:
+        base_name = "sample"
+    
+    quant_column_name = f"quant_{base_name}.csv"
+
+    # Prepare the SLAW dataframe with required columns
+    import polars as pl
+    
+    df = self.features_df
+    
+    # Evaluate the charge column first if adduct_charge exists
+    if "adduct_charge" in df.columns:
+        charge_series = df.select(
+            pl.when(pl.col("adduct_charge") == 0)
+            .then(1 if self.polarity == "positive" else -1)
+            .otherwise(pl.col("adduct_charge"))
+            .alias("charge")
+        ).get_column("charge")
+    else:
+        charge_series = pl.Series([1 if self.polarity == "positive" else -1] * len(df))
+
+    # Evaluate the group column (from adduct_group)
+    # Features with adduct_group == 0 should each get a unique group index
+    if "adduct_group" in df.columns:
+        max_adduct_group = df.get_column("adduct_group").max()
+        if max_adduct_group is None:
+            max_adduct_group = 0
+        
+        # Create a row number starting from max_adduct_group + 1 for features with adduct_group == 0
+        group_series = df.select(
+            pl.when(pl.col("adduct_group") == 0)
+            .then(max_adduct_group + 1 + pl.int_range(pl.len()).over(pl.col("adduct_group") == 0))
+            .otherwise(pl.col("adduct_group"))
+            .alias("group")
+        ).get_column("group")
+    else:
+        group_series = pl.Series([None] * len(df))
+
+    # Evaluate the annotation column (adduct + isotope info)
+    # annotation = adduct for iso==0, adduct + " +{iso}" for iso>0
+    if "adduct" in df.columns and "iso" in df.columns:
+        annotation_series = df.select(
+            pl.when(pl.col("iso") == 0)
+            .then(pl.col("adduct").str.replace(r"\?", "H"))
+            .otherwise(pl.col("adduct").str.replace(r"\?", "H") + " +" + pl.col("iso").cast(pl.Utf8))
+            .alias("annotation")
+        ).get_column("annotation")
+    elif "adduct" in df.columns:
+        annotation_series = df.get_column("adduct").str.replace(r"\?", "H")
+    else:
+        annotation_series = pl.Series([""] * len(df))
+
+    # Create SLAW columns with appropriate mappings from features_df
+    # Columns are ordered according to SLAW specification
+    slaw_data = {
+        "feature_id": df.get_column("feature_id") if "feature_id" in df.columns else pl.Series(range(1, len(df) + 1)),
+        "mz": df.get_column("mz") if "mz" in df.columns else pl.Series([None] * len(df)),
+        "rt": df.get_column("rt") if "rt" in df.columns else pl.Series([None] * len(df)),
+        "group": group_series,
+        "annotation": annotation_series,
+        "neutral_mass": df.get_column("adduct_neutral_mass") if "adduct_neutral_mass" in df.columns else pl.Series([None] * len(df)),
+        "charge": charge_series,
+        "main_id": df.get_column("main_id") if "main_id" in df.columns else df.get_column("feature_id") if "feature_id" in df.columns else pl.Series(range(1, len(df) + 1)),
+        "ion": df.get_column("adduct").str.replace(r"\?", "H") if "adduct" in df.columns else pl.Series([""] * len(df)),
+        "iso": df.get_column("iso") if "iso" in df.columns else pl.Series([0] * len(df)),
+        "clique": df.get_column("clique") if "clique" in df.columns else pl.Series([None] * len(df)),
+        "num_detection": pl.Series([1] * len(df)),  # Single sample always 1
+        "total_detection": pl.Series([1] * len(df)),  # Single sample always 1
+        "mz_mean": df.get_column("mz") if "mz" in df.columns else pl.Series([None] * len(df)),
+        "mz_min": df.get_column("mz") if "mz" in df.columns else pl.Series([None] * len(df)),
+        "mz_max": df.get_column("mz") if "mz" in df.columns else pl.Series([None] * len(df)),
+        "rt_mean": df.get_column("rt") if "rt" in df.columns else pl.Series([None] * len(df)),
+        "rt_min": df.get_column("rt_start") if "rt_start" in df.columns else df.get_column("rt") if "rt" in df.columns else pl.Series([None] * len(df)),
+        "rt_max": df.get_column("rt_end") if "rt_end" in df.columns else df.get_column("rt") if "rt" in df.columns else pl.Series([None] * len(df)),
+        "rt_cor_mean": df.get_column("rt") if "rt" in df.columns else pl.Series([None] * len(df)),
+        "rt_cor_min": df.get_column("rt_start") if "rt_start" in df.columns else df.get_column("rt") if "rt" in df.columns else pl.Series([None] * len(df)),
+        "rt_cor_max": df.get_column("rt_end") if "rt_end" in df.columns else df.get_column("rt") if "rt" in df.columns else pl.Series([None] * len(df)),
+        "height_mean": df.get_column("height") if "height" in df.columns else pl.Series([None] * len(df)),
+        "height_min": df.get_column("height") if "height" in df.columns else pl.Series([None] * len(df)),
+        "height_max": df.get_column("height") if "height" in df.columns else pl.Series([None] * len(df)),
+        "intensity_mean": df.get_column("inty") if "inty" in df.columns else pl.Series([None] * len(df)),
+        "intensity_min": df.get_column("inty") if "inty" in df.columns else pl.Series([None] * len(df)),
+        "intensity_max": df.get_column("inty") if "inty" in df.columns else pl.Series([None] * len(df)),
+        "SN_mean": df.get_column("sn") if "sn" in df.columns else pl.Series([None] * len(df)),
+        "SN_min": df.get_column("sn") if "sn" in df.columns else pl.Series([None] * len(df)),
+        "SN_max": df.get_column("sn") if "sn" in df.columns else pl.Series([None] * len(df)),
+        "peakwidth_mean": (df.get_column("rt_end") - df.get_column("rt_start")) if ("rt_end" in df.columns and "rt_start" in df.columns) else pl.Series([None] * len(df)),
+        "peakwidth_min": (df.get_column("rt_end") - df.get_column("rt_start")) if ("rt_end" in df.columns and "rt_start" in df.columns) else pl.Series([None] * len(df)),
+        "peakwidth_max": (df.get_column("rt_end") - df.get_column("rt_start")) if ("rt_end" in df.columns and "rt_start" in df.columns) else pl.Series([None] * len(df)),
+        "ms2_mgf_id": df.get_column("ms2_mgf_id") if "ms2_mgf_id" in df.columns else pl.Series([""] * len(df)),
+        "ms2_num_fused": df.get_column("ms2_scans").list.len() if "ms2_scans" in df.columns and df["ms2_scans"].dtype == pl.List else pl.Series([None] * len(df)),
+        "ms2_source": df.get_column("ms2_source") if "ms2_source" in df.columns else pl.Series([""] * len(df)),
+        "isotopic_pattern_annot": df.get_column("isotopic_pattern_annot") if "isotopic_pattern_annot" in df.columns else pl.Series([""] * len(df)),
+        "isotopic_pattern_rel": df.get_column("isotopic_pattern_rel") if "isotopic_pattern_rel" in df.columns else pl.Series([""] * len(df)),
+        "isotopic_pattern_abs": df.get_column("isotopic_pattern_abs") if "isotopic_pattern_abs" in df.columns else pl.Series([""] * len(df)),
+        quant_column_name: df.get_column("inty") if "inty" in df.columns else pl.Series([None] * len(df)),
+    }
+    
+    # Create the polars DataFrame
+    slaw_df = pl.DataFrame(slaw_data)
+    
+    # Convert to pandas for CSV export with comma separator
+    pandas_df = slaw_df.to_pandas()
+    
+    # Export to CSV with comma separator - only quote when necessary (QUOTE_MINIMAL)
+    try:
+        pandas_df.to_csv(filename, sep=',', index=False, quoting=0)  # quoting=0 means QUOTE_MINIMAL
+        self.logger.success(f"Features exported to {filename} (SLAW format)")
+        self.logger.debug(f"Exported {len(slaw_df)} features with {len(slaw_df.columns)} columns")
+    except PermissionError:
+        self.logger.error(f"Permission denied: Cannot write to {filename}. The file may be open in another program. Please close it and try again.")
+
+
 def export_chrom(self, filename="chrom.csv"):
     # saves self.chrom_df to a csv file. Remove the scan_uid and chrom columns if the file already exists
     if self.chrom_df is None:
@@ -872,3 +1009,589 @@ def export_chrom(self, filename="chrom.csv"):
     if "chrom" in data.columns:
         data = data.drop("chrom")
     data.to_csv(filename, index=False)
+
+
+def export_mztab(self, filename=None, title=None, description=None, include_mgf=False, **kwargs):
+    """
+    Export the sample as a fully compliant mzTab-M file.
+
+    Args:
+        filename (str, optional): Path to the output mzTab-M file. Defaults to "sample.mztab".
+        title (str, optional): Human-readable title for the file.
+        description (str, optional): Human-readable description.
+        include_mgf (bool, optional): Include MGF table with MS2 spectra. Defaults to False.
+        **kwargs: Additional metadata or export options.
+    """
+    from masster._version import __version__
+
+    def safe_str(value, default="null"):
+        """Convert value to string, replacing empty strings with 'null'"""
+        if value is None:
+            return default
+        str_val = str(value)
+        return str_val if str_val.strip() != "" else default
+
+    if filename is None:
+        filename = "sample.mztab"
+    if not os.path.isabs(filename):
+        filename = os.path.abspath(filename)
+
+    # Get identification data if available using get_id() function
+    id_data = None
+    top_id_data = None
+    full_id_data = None
+    
+    try:
+        # Import get_id function from sample.id module
+        from masster.sample.id import get_id
+        
+        # Get full enriched identification data
+        full_id_data = get_id(self)
+        if full_id_data is not None and not full_id_data.is_empty():
+            # Get top scoring identification for each feature_uid for SML section
+            top_id_data = (
+                full_id_data.group_by("feature_uid")
+                .agg(pl.all().sort_by("score", descending=True).first())
+                .sort("feature_uid")
+            )
+            # Keep raw id_data for backward compatibility (if needed elsewhere)
+            id_data = self.id_df if hasattr(self, "id_df") and self.id_df is not None else None
+        else:
+            self.logger.info("No identification data available for mzTab export")
+    except Exception as e:
+        self.logger.debug(f"Could not retrieve identification data: {e}")
+        id_data = None
+        top_id_data = None
+        full_id_data = None
+
+    # Get MGF data only if requested
+    mgf_data = None
+    mgf_mapping: dict[int, list[int]] = {}
+    if include_mgf:
+        # Create MGF data from features_df
+        if self.features_df is not None:
+            mgf_rows = []
+            mgf_index = 1
+            
+            for feature_row in self.features_df.iter_rows(named=True):
+                feature_uid = feature_row["feature_uid"]
+                feature_id = feature_row.get("feature_id", feature_uid)
+                
+                # Check if this feature has MS2 scans
+                if feature_row.get("ms2_scans") is None:
+                    continue
+                
+                ms2_scans = feature_row["ms2_scans"]
+                if not isinstance(ms2_scans, list):
+                    ms2_scans = [ms2_scans]
+                
+                # Process each MS2 scan
+                for scan_uid in ms2_scans:
+                    spec = self.get_spectrum(scan_uid)
+                    if spec is None or len(spec.mz) == 0:
+                        continue
+                    
+                    mgf_row = {
+                        "mgf_index": mgf_index,
+                        "feature_uid": feature_uid,
+                        "feature_id": feature_id,
+                        "rtinseconds": feature_row.get("rt", 0),
+                        "pepmass": feature_row.get("mz", 0),
+                        "energy": spec.energy if hasattr(spec, "energy") else 0,
+                        "mslevel": spec.ms_level if hasattr(spec, "ms_level") else 2,
+                        "title": f"uid:{feature_uid}, rt:{feature_row.get('rt', 0):.2f}, mz:{feature_row.get('mz', 0):.4f}",
+                        "spec_mz": spec.mz,
+                        "spec_int": spec.inty,
+                        "spec_len": len(spec.mz),
+                    }
+                    mgf_rows.append(mgf_row)
+                    
+                    # Track mapping
+                    if feature_uid not in mgf_mapping:
+                        mgf_mapping[feature_uid] = []
+                    mgf_mapping[feature_uid].append(mgf_index)
+                    
+                    mgf_index += 1
+            
+            if mgf_rows:
+                mgf_data = pl.DataFrame(mgf_rows)
+
+    # --- Prepare MTD (metadata) section ---
+    mtd_lines = []
+    mtd_lines.append(
+        f"COM\tfile generated by MASSter {__version__} on {datetime.now().strftime('%Y-%m-%d %H:%M:%S')}",
+    )
+    mtd_lines.append("\nMTD\tmzTab-version\t2.2.0-M")
+    
+    # Use sample name or filename as mzTab-id
+    sample_name = getattr(self, "label", None) or os.path.splitext(os.path.basename(self.file_path))[0] if hasattr(self, "file_path") and self.file_path else "sample"
+    mtd_lines.append(f"MTD\tmzTab-id\t{sample_name}")
+    mtd_lines.append("")
+    
+    # CV definitions
+    mtd_lines.append("MTD\tcv[1]-label\tMS")
+    mtd_lines.append("MTD\tcv[1]-full_name\tPSI-MS controlled vocabulary")
+    mtd_lines.append("MTD\tcv[1]-version\t4.1.199")
+    mtd_lines.append(
+        "MTD\tcv[1]-uri\thttps://raw.githubusercontent.com/HUPO-PSI/psi-ms-CV/master/psi-ms.obo",
+    )
+    mtd_lines.append("")
+    
+    # Quantification units
+    mtd_lines.append(
+        "MTD\tsmall_molecule-quantification_unit\t[MS, MS:1001844, MS1 feature area, ]",
+    )
+    mtd_lines.append(
+        "MTD\tsmall_molecule_feature-quantification_unit\t[MS, MS:1001844, MS1 feature area, ]",
+    )
+    mtd_lines.append(
+        "MTD\tsmall_molecule-identification_reliability\t[MS, MS:1002955, hr-ms compound identification confidence level, ]",
+    )
+    
+    # Identification confidence
+    mtd_lines.append(
+        "MTD\tid_confidence_measure[1]\t[MS, MS:1002888, small molecule confidence measure, ]",
+    )
+    mtd_lines.append("")
+    
+    # Software
+    mtd_lines.append("MTD\tsoftware[1]\t[MS, MS:1003430, OpenMS, unknown]")
+    mtd_lines.append(f"MTD\tsoftware[2]\t[MS, MS:1002878, MASSter, {__version__}]")
+    mtd_lines.append(
+        "MTD\tquantification_method\t[MS, MS:1001834, LC-MS label-free quantitation analysis, ]",
+    )
+    mtd_lines.append("")
+    
+    # Database information - updated based on identification data
+    if full_id_data is not None and hasattr(self, "lib_df") and self.lib_df is not None and not self.lib_df.is_empty():
+        mtd_lines.append('MTD\tdatabase[1]\t[, , "compound library", ]')
+        mtd_lines.append("MTD\tdatabase[1]-prefix\tcmpd")
+        mtd_lines.append("MTD\tdatabase[1]-version\tUnknown")
+        mtd_lines.append("MTD\tdatabase[1]-uri\thttps://pubchem.ncbi.nlm.nih.gov/")
+    else:
+        mtd_lines.append('MTD\tdatabase[1]\t[, , "PubChem", ]')
+        mtd_lines.append("MTD\tdatabase[1]-prefix\tCID")
+        mtd_lines.append("MTD\tdatabase[1]-version\tUnknown")
+        mtd_lines.append("MTD\tdatabase[1]-uri\thttps://pubchem.ncbi.nlm.nih.gov/")
+    
+    # Single sample metadata
+    mtd_lines.append(f"\nMTD\tsample[1]\t{sample_name}")
+    mtd_lines.append(f"MTD\tsample[1]-description\t{sample_name}")
+    mtd_lines.append(f"MTD\tms_run[1]-location\tfile://unknown")
+    
+    # Scan polarity
+    sample_polarity = getattr(self, "polarity", "positive")
+    if sample_polarity in ["negative", "neg"]:
+        scan_polarity_cv = "[MS, MS:1000129, negative scan, ]"
+    else:
+        scan_polarity_cv = "[MS, MS:1000130, positive scan, ]"
+    mtd_lines.append(f"MTD\tms_run[1]-scan_polarity\t{scan_polarity_cv}")
+    
+    mtd_lines.append("MTD\tassay[1]\tAssay_1")
+    mtd_lines.append("MTD\tassay[1]-sample_ref\tsample[1]")
+    mtd_lines.append("MTD\tassay[1]-ms_run_ref\tms_run[1]")
+    mtd_lines.append("")
+    mtd_lines.append("MTD\tstudy_variable[1]\tundefined")
+    mtd_lines.append("MTD\tstudy_variable[1]-assay_refs\tassay[1]")
+    mtd_lines.append("MTD\tstudy_variable[1]-description\tSingle sample")
+    
+    with open(filename, "w", encoding="utf-8") as f:
+        for line in mtd_lines:
+            f.write(line + "\n")
+
+    # --- SML (Small Molecule) table ---
+    sml_lines = []
+    sml_header = [
+        "SMH",
+        "SML_ID",
+        "SMF_ID_REFS",
+        "database_identifier",
+        "chemical_formula",
+        "smiles",
+        "inchi",
+        "chemical_name",
+        "uri",
+        "theoretical_neutral_mass",
+        "adduct_ions",
+        "reliability",
+        "best_id_confidence_measure",
+        "best_id_confidence_value",
+        "opt_global_mgf_index",
+        "abundance_assay[1]",
+        "abundance_study_variable[1]",
+        "abundance_variation_study_variable[1]",
+    ]
+    sml_lines.append("\t".join(sml_header))
+    
+    # Get adducts from features_df['adduct']
+    adduct_list = []
+    for row in self.features_df.iter_rows(named=True):
+        adduct = "null"
+        if "adduct" in row and row["adduct"] is not None:
+            adduct = str(row["adduct"]).replace("?", "H")
+        adduct_list.append(adduct)
+    
+    for idx, row in enumerate(self.features_df.iter_rows(named=True), 1):
+        feature_uid = row["feature_uid"]
+        
+        # Get identification information for this feature_uid if available
+        id_info = None
+        if top_id_data is not None:
+            id_matches = top_id_data.filter(pl.col("feature_uid") == feature_uid)
+            if id_matches.height > 0:
+                id_info = id_matches.row(0, named=True)
+        
+        # Populate identification fields
+        database_identifier = "null"
+        chemical_formula = "null"
+        smiles_val = "null"
+        inchi_val = "null"
+        chemical_name = "null"
+        best_id_confidence_measure = "null"
+        best_id_confidence_value = "null"
+        reliability = "4"  # Default: unknown compound
+        theoretical_neutral_mass = "null"
+        
+        if id_info:
+            # Use cmpd_uid as database identifier with prefix
+            if id_info.get("cmpd_uid") is not None:
+                database_identifier = f"cmpd:{id_info['cmpd_uid']}"
+            
+            # Chemical formula
+            if id_info.get("formula") is not None and id_info["formula"] != "":
+                chemical_formula = safe_str(id_info["formula"])
+            
+            # SMILES
+            if id_info.get("smiles") is not None and id_info["smiles"] != "":
+                smiles_val = safe_str(id_info["smiles"])
+            
+            # InChI
+            if id_info.get("inchi") is not None and id_info["inchi"] != "":
+                inchi_val = safe_str(id_info["inchi"])
+            
+            # Chemical name
+            if id_info.get("name") is not None and id_info["name"] != "":
+                chemical_name = safe_str(id_info["name"])
+            
+            # Theoretical neutral mass
+            if id_info.get("neutral_mass") is not None:
+                theoretical_neutral_mass = safe_str(id_info["neutral_mass"])
+            elif id_info.get("mass") is not None:
+                theoretical_neutral_mass = safe_str(id_info["mass"])
+            
+            # Identification confidence
+            if id_info.get("matcher") is not None:
+                best_id_confidence_measure = f"[MS, MS:1002888, {id_info['matcher']}, ]"
+            
+            if id_info.get("score") is not None:
+                best_id_confidence_value = safe_str(id_info["score"])
+            
+            # Set reliability based on identification quality
+            if id_info.get("score", 0) >= 0.8:
+                reliability = "2a"  # High confidence compound match
+            elif id_info.get("score", 0) >= 0.5:
+                reliability = "2b"  # Moderate confidence match
+            elif id_info.get("score", 0) >= 0.2:
+                reliability = "3"  # Compound class level
+            else:
+                reliability = "4"  # Unknown compound
+        
+        # Get MGF indexes for this feature
+        mgf_indexes = mgf_mapping.get(feature_uid, [])
+        
+        # Get intensity value for abundance
+        abundance_value = row.get("inty", None)
+        abundance_str = safe_str(abundance_value) if abundance_value is not None else "null"
+        
+        sml_row = [
+            "SML",
+            str(idx),
+            str(idx),  # SMF_ID_REFS - same as SML_ID for single features
+            database_identifier,
+            chemical_formula,
+            smiles_val,
+            inchi_val,
+            chemical_name,
+            safe_str(row.get("uri", "null")),
+            theoretical_neutral_mass,
+            adduct_list[idx - 1],
+            reliability,
+            best_id_confidence_measure,
+            best_id_confidence_value,
+            ",".join(map(str, mgf_indexes)) if mgf_indexes else "null",
+            abundance_str,  # abundance_assay[1]
+            abundance_str,  # abundance_study_variable[1] (same for single sample)
+            "null",  # abundance_variation_study_variable[1] (no variation for single sample)
+        ]
+        sml_lines.append("\t".join(sml_row))
+    
+    with open(filename, "a", encoding="utf-8") as f:
+        f.write("\n")
+        for line in sml_lines:
+            f.write(line + "\n")
+
+    # --- SMF (Small Molecule Feature) table ---
+    smf_lines = []
+    smf_header = [
+        "SFH",
+        "SMF_ID",
+        "SME_ID_REFS",
+        "SME_ID_REF_ambiguity_code",
+        "adduct_ion",
+        "isotopomer",
+        "exp_mass_to_charge",
+        "charge",
+        "retention_time_in_seconds",
+        "retention_time_in_seconds_start",
+        "retention_time_in_seconds_end",
+        "abundance_assay[1]",
+        "abundance_study_variable[1]",
+        "abundance_variation_study_variable[1]",
+    ]
+    smf_lines.append("\t".join(smf_header))
+    
+    for idx, row in enumerate(self.features_df.iter_rows(named=True), 1):
+        feature_uid = row["feature_uid"]
+        
+        # References to SME entries
+        SME_refs = "null"
+        SME_ambiguity = "null"
+        
+        if full_id_data is not None:
+            # Find all SME entries for this feature_uid
+            SME_matches = full_id_data.filter(pl.col("feature_uid") == feature_uid)
+            if SME_matches.height > 0:
+                # Generate SME IDs
+                SME_ids = []
+                for i, SME_row in enumerate(SME_matches.iter_rows(named=True)):
+                    SME_id_base = feature_uid * 1000
+                    SME_id = SME_id_base + i + 1
+                    SME_ids.append(str(SME_id))
+                
+                if SME_ids:
+                    SME_refs = "|".join(SME_ids)
+                    # Set ambiguity code
+                    if len(SME_ids) > 1:
+                        unique_cmpds = {
+                            match["cmpd_uid"]
+                            for match in SME_matches.iter_rows(named=True)
+                            if match.get("cmpd_uid") is not None
+                        }
+                        if len(unique_cmpds) > 1:
+                            SME_ambiguity = "1"  # Ambiguous identification
+                        else:
+                            SME_ambiguity = "2"  # Multiple evidence for same molecule
+                    else:
+                        SME_ambiguity = "null"
+        
+        # Format isotopomer
+        iso_value = row.get("iso", 0)
+        if iso_value is not None and round(iso_value) != 0:
+            isotopomer = f'[MS,MS:1002957,"isotopomer MS peak","+{round(iso_value)}"]'
+        else:
+            isotopomer = "null"
+        
+        # Get abundance value
+        abundance_value = row.get("inty", None)
+        abundance_str = safe_str(abundance_value) if abundance_value is not None else "null"
+        
+        smf_row = [
+            "SMF",
+            str(idx),
+            SME_refs,
+            SME_ambiguity,
+            adduct_list[idx - 1],  # adduct_ion
+            isotopomer,
+            safe_str(row.get("mz", "null")),  # exp_mass_to_charge
+            safe_str(row.get("charge", "null")),
+            safe_str(row.get("rt", "null")),  # retention_time_in_seconds
+            safe_str(row.get("rt_start", "null")),
+            safe_str(row.get("rt_end", "null")),
+            abundance_str,  # abundance_assay[1]
+            abundance_str,  # abundance_study_variable[1]
+            "null",  # abundance_variation_study_variable[1]
+        ]
+        smf_lines.append("\t".join(smf_row))
+    
+    with open(filename, "a", encoding="utf-8") as f:
+        f.write("\n")
+        for line in smf_lines:
+            f.write(line + "\n")
+
+    # --- SME (Small Molecule Evidence) table ---
+    if full_id_data is not None and not full_id_data.is_empty():
+        SME_lines = []
+        SME_lines.append(
+            "COM\tThe spectra_ref are dummy placeholders, as the annotation was based on aggregated data",
+        )
+        SME_header = [
+            "SEH",
+            "SME_ID",
+            "evidence_input_id",
+            "database_identifier",
+            "chemical_formula",
+            "smiles",
+            "inchi",
+            "chemical_name",
+            "uri",
+            "derivatized_form",
+            "adduct_ion",
+            "exp_mass_to_charge",
+            "charge",
+            "theoretical_mass_to_charge",
+            "spectra_ref",
+            "identification_method",
+            "ms_level",
+            "id_confidence_measure[1]",
+            "rank",
+        ]
+        SME_lines.append("\t".join(SME_header))
+        
+        # Create SME entries for all identification results
+        for feature_uid in self.features_df.select("feature_uid").to_series().unique():
+            # Get feature data
+            feature_data = self.features_df.filter(pl.col("feature_uid") == feature_uid)
+            if feature_data.height == 0:
+                continue
+            feature_row = feature_data.row(0, named=True)
+            
+            # Get all identification results for this feature
+            SME_matches = full_id_data.filter(pl.col("feature_uid") == feature_uid)
+            
+            if SME_matches.height > 0:
+                # Sort by score descending
+                SME_matches = SME_matches.sort("score", descending=True)
+                
+                for i, SME_row in enumerate(SME_matches.iter_rows(named=True)):
+                    # Generate unique SME_ID
+                    SME_id_base = feature_uid * 1000
+                    SME_id = SME_id_base + i + 1
+                    
+                    # Create evidence input ID
+                    feature_mz = feature_row.get("mz", 0)
+                    feature_rt = feature_row.get("rt", 0)
+                    feature_id = feature_row.get("feature_id", feature_uid)
+                    evidence_id = f"feature_uid={feature_uid}:feature_id={feature_id}:mz={feature_mz:.4f}:rt={feature_rt:.2f}"
+                    
+                    # Database identifier
+                    db_id = "null"
+                    if SME_row.get("db_id") is not None and SME_row["db_id"] != "":
+                        db_id = safe_str(SME_row["db_id"])
+                    elif SME_row.get("cmpd_uid") is not None:
+                        db_id = f"cmpd:{SME_row['cmpd_uid']}"
+                    
+                    # Get adduct information
+                    adduct_ion = "null"
+                    if SME_row.get("adduct") is not None and SME_row["adduct"] != "":
+                        adduct_ion = safe_str(SME_row["adduct"]).replace("?", "H")
+                    
+                    # Spectra reference
+                    spectra_ref = "ms_run[1]:spectrum=0"
+                    
+                    # Identification method
+                    id_method = "[MS, MS:1002888, small molecule confidence measure, ]"
+                    if SME_row.get("matcher") is not None:
+                        id_method = f"[MS, MS:1002888, {SME_row['matcher']}, ]"
+                    
+                    # MS level - check if ms1 exists in matched
+                    if 'ms1' in SME_row['matcher'].lower():                        
+                        ms_level = "[MS, MS:1000511, ms level, 1]"
+                    else:
+                        ms_level = "[MS,MS:1000511, ms level, 2]"
+                    
+                    # Experimental mass-to-charge
+                    exp_mz = safe_str(feature_mz)
+                    
+                    # Theoretical mass-to-charge
+                    theoretical_mz = "null"
+                    if SME_row.get("mz") is not None:
+                        theoretical_mz = safe_str(SME_row["mz"])
+                    
+                    SME_line = [
+                        "SME",
+                        str(SME_id),
+                        evidence_id,
+                        db_id,
+                        safe_str(SME_row.get("formula", "null")),
+                        safe_str(SME_row.get("smiles", "null")),
+                        safe_str(SME_row.get("inchi", "null")),
+                        safe_str(SME_row.get("name", "null")),
+                        "null",  # uri
+                        "null",  # derivatized_form
+                        adduct_ion,
+                        exp_mz,
+                        safe_str(feature_row.get("charge", "1")),
+                        theoretical_mz,
+                        spectra_ref,
+                        id_method,
+                        ms_level,
+                        safe_str(SME_row.get("score", "null")),
+                        str(i + 1),  # rank
+                    ]
+                    SME_lines.append("\t".join(SME_line))
+        
+        # Write SME table
+        with open(filename, "a", encoding="utf-8") as f:
+            f.write("\n")
+            for line in SME_lines:
+                f.write(line + "\n")
+
+    # --- MGF table ---
+    if include_mgf and mgf_data is not None and len(mgf_data) > 0:
+        mgf_lines = []
+        # Header
+        mgf_header = [
+            "COM",
+            "MGH",
+            "mgf_id",
+            "prec_id",
+            "prec_rt",
+            "prec_mz",
+            "prec_int",
+            "energy",
+            "level",
+            "title",
+            "spec_tic",
+            "spec_len",
+            "spec_mz",
+            "spec_int",
+        ]
+        mgf_lines.append("\t".join(mgf_header))
+        
+        # Data rows
+        for row in mgf_data.iter_rows(named=True):
+            # Calculate spectrum TIC
+            spectrum_mz = row["spec_mz"]
+            spectrum_inty = row["spec_int"]
+            spec_tic = sum(spectrum_inty) if spectrum_inty else 0
+            spec_len = row["spec_len"] if row["spec_len"] is not None else 0
+            
+            # Format spectrum data as pipe-separated strings
+            spec_mz_str = "|".join([f"{mz:.4f}" for mz in spectrum_mz]) if spectrum_mz else ""
+            spec_int_str = "|".join([f"{int(inty)}" for inty in spectrum_inty]) if spectrum_inty else ""
+            
+            mgf_row = [
+                "COM",
+                "MGF",
+                str(row["mgf_index"]) if row["mgf_index"] is not None else "null",
+                str(row["feature_id"]) if row["feature_id"] is not None else "null",
+                f"{row['rtinseconds']:.2f}" if row["rtinseconds"] is not None else "null",
+                f"{row['pepmass']:.4f}" if row["pepmass"] is not None else "null",
+                "null",  # prec_int
+                str(row["energy"]) if row["energy"] is not None else "null",
+                str(row["mslevel"]) if row["mslevel"] is not None else "null",
+                str(row["title"]) if row["title"] is not None else "null",
+                f"{int(spec_tic)}" if spec_tic > 0 else "null",
+                str(spec_len) if spec_len > 0 else "null",
+                spec_mz_str if spec_mz_str else "null",
+                spec_int_str if spec_int_str else "null",
+            ]
+            mgf_lines.append("\t".join(mgf_row))
+        
+        # Write MGF table
+        with open(filename, "a", encoding="utf-8") as f:
+            f.write("\n")
+            for line in mgf_lines:
+                f.write(line + "\n")
+
+    self.logger.success(f"Exported mzTab-M to {filename}")