ms-mint 0.2.3__tar.gz → 0.2.4__tar.gz

{ms-mint-0.2.3/ms_mint.egg-info → ms-mint-0.2.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: ms-mint
-Version: 0.2.3
+Version: 0.2.4
 Summary: Metabolomics Integrator (Mint)
 Home-page: https://github.com/LewisResearchGroup/ms-mint
 Author: Soren Wacker

{ms-mint-0.2.3 → ms-mint-0.2.4}/ms_mint/_version.py

@@ -8,11 +8,11 @@ import json
 
 version_json = '''
 {
- "date": "2023-04-26T16:56:04-0600",
+ "date": "2023-05-09T10:35:26-0600",
  "dirty": false,
  "error": null,
- "full-revisionid": "087cd80fb933baa6736368cb6d00e0883d26a632",
- "version": "v0.2.3"
+ "full-revisionid": "509e8d0641ce628f2c4e4fdf65a850d805427c99",
+ "version": "v0.2.4"
 }
 '''  # END VERSION_JSON
 

{ms-mint-0.2.3 → ms-mint-0.2.4}/ms_mint/io.py

@@ -16,10 +16,9 @@ from pyteomics import mzxml, mzml
 
 try:
     from pyteomics import mzmlb
-
     MZMLB_AVAILABLE = True
-except ImportError:
-    logging.warning("Cound not import pyteomics.mzmlb")
+except ImportError as e:
+    logging.warning(f"Cound not import pyteomics.mzmlb:\n{e}")
     MZMLB_AVAILABLE = False
 
 
@@ -300,7 +299,7 @@ def mzmlb_to_df__pyteomics(fn, read_only=False):
     data = list(extract_mzmlb(data))
     df = (
         pd.DataFrame.from_dict(data)
-        .set_index(["index", "retentionTime"])
+        .set_index(["index", "retentionTime", "polarity"])
         .apply(pd.Series.explode)
         .reset_index()
         .rename(
@@ -316,14 +315,20 @@ def mzmlb_to_df__pyteomics(fn, read_only=False):
 
     # mzMLb starts scan index with 0
     df["scan_id"] = df["scan_id"] + 1
-    df["polarity"] = None
+    
     df = df[MS_FILE_COLUMNS]
     return df
 
 
 def _extract_mzmlb(data):
-    cols = ["index", "ms level", "retentionTime", "m/z array", "intensity array"]
+    cols = ["index", "ms level", "polarity", "retentionTime", "m/z array", "intensity array"]
     data["retentionTime"] = data["scanList"]["scan"][0]["scan start time"] * 60
+    if 'positive scan' in data.keys():
+        data["polarity"] = '+'
+    elif 'negative scan' in data.keys():
+        data["polarity"] = '-'
+    else:
+        data["polarity"] = None
     return {c: data[c] for c in cols}
 
 

{ms-mint-0.2.3 → ms-mint-0.2.4}/ms_mint/processing.py

@@ -50,8 +50,6 @@ def process_ms1_files_in_parallel(args):
         logging.error(f"process_ms1_files_in_parallel(): {e}")
         results = pd.DataFrame()
 
-    print(len(results))
-
     if (output_fn is not None) and (len(results) > 0):
         append_results(results, output_fn)
         return None
@@ -90,7 +88,6 @@ def process_ms1_file(filename, targets):
     results["ms_path"] = os.path.dirname(filename)
     results["ms_file_size"] = os.path.getsize(filename) / 1024 / 1024
     results["peak_score"] = score_peaks(results)
-    print(results)
     return results[MINT_RESULTS_COLUMNS]
 
 
@@ -191,14 +188,6 @@ def extract_ms1_properties(array, mz_mean):
 
     projection = pd.DataFrame(array[:, [0, 2]], columns=["rt", "int"])
 
-    print("DEBUG extract_ms1_properties")
-    print(array)
-    print(array.dtype)
-    print(array.shape)
-    print(len(projection))
-    print(projection)
-    print(projection.dtypes)
-
     projection["rt"] = projection["rt"].round(2)
     projection["int"] = projection["int"].astype(int)
     projection = projection.groupby("rt").max().reset_index().values

{ms-mint-0.2.3 → ms-mint-0.2.4/ms_mint.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: ms-mint
-Version: 0.2.3
+Version: 0.2.4
 Summary: Metabolomics Integrator (Mint)
 Home-page: https://github.com/LewisResearchGroup/ms-mint
 Author: Soren Wacker

{ms-mint-0.2.3 → ms-mint-0.2.4}/ms_mint.egg-info/SOURCES.txt

@@ -24,4 +24,5 @@ ms_mint.egg-info/PKG-INFO
 ms_mint.egg-info/SOURCES.txt
 ms_mint.egg-info/dependency_links.txt
 ms_mint.egg-info/requires.txt
-ms_mint.egg-info/top_level.txt
+ms_mint.egg-info/top_level.txt
+scripts/ms-mint-convert.py

{ms-mint-0.2.3 → ms-mint-0.2.4}/ms_mint.egg-info/requires.txt

@@ -21,3 +21,5 @@ ipyfilechooser
 openpyxl
 pyarrow
 tables
+h5py
+hdf5plugin

ms-mint-0.2.4/scripts/ms-mint-convert.py

@@ -0,0 +1,43 @@
+#!/usr/bin/env python
+
+from ms_mint import io
+from pathlib import Path as P
+from tqdm import tqdm
+import argparse
+import os
+import logging
+
+
+def convert(fn, fmt='parquet', output_directory=None):
+    
+    if output_directory is None:
+        output_directory = P(fn).parent
+        
+    fn_out = output_directory / P(fn).with_suffix(f".{fmt}")
+    
+    if fn_out.is_file():
+        logging.error(f"File exists {fn_out}")
+    else:
+        os.makedirs(output_directory, exist_ok=True)
+        logging.info(f"{fn} --> {fn_out}")
+        
+        if fmt == 'parquet':
+          io.ms_file_to_df(fn).to_parquet(fn_out)
+        elif fmt == 'feather':
+          io.ms_file_to_df(fn).to_feather(fn_out)
+         
+        
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+    parser.add_argument("-i", "--input", nargs="+", required=True)
+    parser.add_argument("-o", "--output-directory")
+    parser.add_argument("-f", "--format", choices=['parquet', 'feather'], default='parquet')
+
+    args = parser.parse_args()
+    fns = args.input
+    output_directory = args.output_directory
+    fmt = args.format
+    
+    for fn in tqdm(fns):
+        convert(fn, fmt, output_directory)

{ms-mint-0.2.3 → ms-mint-0.2.4}/setup.py

@@ -31,9 +31,15 @@ install_reqs = [
     "openpyxl",
     "pyarrow",
     "tables",
+    "h5py",
+    "hdf5plugin"
 ]
 
 
+scripts = [
+    "scripts/ms-mint-convert.py"
+]
+
 config = {
     "name": "ms-mint",
     "version": versioneer.get_version(),
@@ -53,6 +59,7 @@ config = {
     "python_requires": ">=3.8",
     "install_requires": install_reqs,
     "include_package_data": True,
+    "scripts": scripts
 }