Merge pull request #35 from ArthurDeclercq/feature/Peaks_id_file_parser

Feature/peaks id file parser

README.md

@@ -48,12 +48,13 @@ MS²Rescore uses identifications from a
 or from the output of one of these search engines:
 
 - [MaxQuant](https://www.maxquant.org/): Start from `msms.txt` identification
-  file and directory with `.mgf` files. (Be sure to export without FDR
-  filtering!)
+  file and directory with `.mgf` files.
 - [MSGFPlus](https://omics.pnl.gov/software/ms-gf): Start with an `.mzid`
   identification file and corresponding `.mgf`.
 - [X!Tandem](https://www.thegpm.org/tandem/): Start with an X!Tandem `.xml`
   identification file and corresponding `.mgf`.
+- [PEAKS](https://www.bioinfor.com/peaksdb/): Start with an `.mzid` identification
+  file and directory with `.mgf` files.
 - [PeptideShaker](http://compomics.github.io/projects/peptide-shaker): Start with a
   PeptideShaker Extended PSM Report and corresponding `.mgf` file.
 
@@ -187,6 +188,17 @@ one of the modifications listed in the MS²PIP configuration (see
     - `fixed_modifications`: Must list all modifications set as fixed during the
 MaxQuant search (as this is not denoted in the msms.txt file). Keys refer to the
 amino acid, values to the modification name used in the MS²PIP configuration.
+    - The maxquant specific configuration could for example be:
+      ```json
+      "maxquant_to_rescore": {
+        "modification_mapping":{
+          "ox":"Oxidation",
+          "cm":"Carbamidomethyl"
+        },
+        "fixed_modifications":{
+          "C":"Carbamidomethyl"
+        }
+      ```
 
 As a general rule, MS²Rescore always needs access to all target and decoy PSMs, not
 only the FDR-filtered targets.

configuration.md

@@ -25,7 +25,6 @@
     - **Items**: Refer to *#/definitions/modifications*.
 - **`percolator`** *(object)*: Command line options directly passed to Percolator (see the Percolator wiki).
 - **`maxquant_to_rescore`** *(object)*: Settings specific to the MaxQuant pipeline. Cannot contain additional properties.
-  - **`mgf_dir`** *(string)*: Path to directory with MGF files.
   - **`modification_mapping`** *(object)*: Mapping of MaxQuant modification labels to modifications names for MS²PIP. Default: `{}`.
   - **`fixed_modifications`** *(object)*: Mapping of amino acids with fixed modifications to the modification name. Default: `{}`.
 ## Definitions

ms2rescore/__init__.py

@@ -8,6 +8,8 @@
 from multiprocessing import cpu_count
 from typing import Dict, Optional, Union
 
+from pandas.errors import EmptyDataError
+
 from ms2pip.ms2pipC import MS2PIP
 
 from ms2rescore import id_file_parser, plotting, rescore_core, setup_logging
@@ -44,6 +46,7 @@ def __init__(
         self.config = parse_config(
             parse_cli_args=parse_cli_args, config_class=configuration
         )
+
         if set_logger:
             setup_logging.setup_logging(self.config["general"]["log_level"])
 
@@ -119,6 +122,8 @@ def _select_pipeline(self):
             pipeline = id_file_parser.TandemPipeline
         elif self.config["general"]["pipeline"] == "peptideshaker":
             pipeline = id_file_parser.PeptideShakerPipeline
+        elif self.config["general"]["pipeline"] == "peaks":
+            pipeline = id_file_parser.PeaksPipeline
         else:
             raise NotImplementedError(self.config["general"]["pipeline"])
         return pipeline
@@ -216,6 +221,7 @@ def _run_percolator(self):
     def run(self):
         """Run MS²ReScore."""
         peprec = self.pipeline.get_peprec()
+        peprec.validate_decoy_presence()
         peprec_filename = self.tmpfile_basepath + ".peprec"
         peprec.to_csv(peprec_filename)
 
@@ -254,35 +260,42 @@ def run(self):
         if self.config["general"]["run_percolator"]:
             self._run_percolator()
 
-        logger.info("Generating Rescore plots")
-        if self.config["general"]["plotting"]:
-
-            plotting.PIN(
-                peprec_filename, self.config["general"]["output_filename"]
-            )
+            # Only use plotting module when run_percolator is true
+            if self.config["general"]["plotting"]:
+                logger.info("Generating Rescore plots")
 
-            for fset in self.config["general"]["feature_sets"]:
-                pout_file = (
-                    self.config["general"]["output_filename"]
-                    + "_"
-                    + "_".join(fset)
-                    + "_features.pout"
+                plotting.PIN(
+                    peprec_filename, self.config["general"]["output_filename"]
                 )
-                pout_decoy_file = (
-                    self.config["general"]["output_filename"]
-                    + "_"
-                    + "_".join(fset)
-                    + "_features.pout_dec"
-                )
-                plotting.POUT(
-                    pout_file,
-                    pout_decoy_file,
-                    self.config["general"]["output_filename"],
-                    " ".join(fset)
-                )
-            plotting.RescoreRecord.save_plots_to_pdf(
-                self.config["general"]["output_filename"] + "_plots.pdf",
-                FDR_thresholds=[0.01, 0.001],
-            )
 
+                for fset in self.config["general"]["feature_sets"]:
+                    pout_file = (
+                        self.config["general"]["output_filename"]
+                        + "_"
+                        + "_".join(fset)
+                        + "_features.pout"
+                    )
+                    pout_decoy_file = (
+                        self.config["general"]["output_filename"]
+                        + "_"
+                        + "_".join(fset)
+                        + "_features.pout_dec"
+                    )
+                    try:
+                        plotting.POUT(
+                            pout_file,
+                            pout_decoy_file,
+                            self.config["general"]["output_filename"],
+                            " ".join(fset)
+                        )
+                    except EmptyDataError:
+                        logger.warn(f"Feature set: {'_'.join(fset)} returned empty pout file")
+                        continue
+
+                plotting.RescoreRecord.save_plots_to_pdf(
+                    self.config["general"]["output_filename"] + "_plots.pdf",
+                    FDR_thresholds=[0.01, 0.001],
+                )
+        if not self.config["general"]["run_percolator"] and self.config["general"]["plotting"]:
+            logger.warn("To plot rescore results run_percolator should be enabled")
         logger.info("MS²ReScore finished!")

ms2rescore/config_parser.py

@@ -72,6 +72,42 @@ def _parse_arguments() -> argparse.Namespace:
         default="info",
         help="logging level (default: `info`)",
     )
+    parser.add_argument(
+        "-n",
+        metavar="VALUE",
+        action="store",
+        type=int,
+        dest="num_cpu",
+        default=None,
+        help="number of cpus available to MS²Rescore",
+    )
+    parser.add_argument(
+        "--pipeline",
+        metavar="PIPELINE",
+        action="store",
+        type=str,
+        dest="pipeline",
+        default=None,
+        help="determines which pipeline to use (default: 'infer')",
+    )
+    parser.add_argument(
+        "--percolator",
+        metavar="BOOL",
+        action="store",
+        type=bool,
+        dest="run_percolator",
+        default=None,
+        help="run percolator (default: 'True')",
+    )
+    parser.add_argument(
+        "--plotting",
+        metavar="BOOL",
+        action="store",
+        type=bool,
+        dest="plotting",
+        default=None,
+        help="write PDF with result plots",
+    )
     return parser.parse_args()
 
 
@@ -156,4 +192,6 @@ def parse_config(parse_cli_args: bool = True, config_class: Optional[Dict] = Non
     config = _validate_filenames(config)
     config = _validate_num_cpu(config)
 
+    config["general"]["pipeline"] = config["general"]["pipeline"].lower()
+
     return config