Add Casanovo-DB Functionality

casanovo/casanovo.py

@@ -62,8 +62,9 @@ def __init__(self, *args, **kwargs) -> None:
             click.Option(
                 ("-m", "--model"),
                 help="""
-                Either the model weights (.ckpt file) or a URL pointing to the model weights
-                file. If not provided, Casanovo will try to download the latest release automatically.
+                Either the model weights (.ckpt file) or a URL pointing to 
+                the model weights file. If not provided, 
+                Casanovo will try to download the latest release automatically.
                 """,
             ),
             click.Option(
@@ -201,6 +202,64 @@ def sequence(
         )
 
 
+@main.command(cls=_SharedParams)
+@click.argument(
+    "peak_path",
+    required=True,
+    nargs=-1,
+    type=click.Path(exists=True, dir_okay=False),
+)
+@click.argument(
+    "fasta_path",
+    required=True,
+    nargs=1,
+    type=click.Path(exists=True, dir_okay=False),
+)
+def db_search(
+    peak_path: Tuple[str],
+    fasta_path: str,
+    model: Optional[str],
+    config: Optional[str],
+    output_dir: Optional[str],
+    output_root: Optional[str],
+    verbosity: str,
+    force_overwrite: bool,
+) -> None:
+    """Perform a database search on MS/MS data using Casanovo-DB.
+
+    PEAK_PATH must be one or more mzML, mzXML, or MGF files.
+    FASTA_PATH must be one FASTA file.
+    """
+    output_path, output_root_name = _setup_output(
+        output_dir, output_root, force_overwrite, verbosity
+    )
+    utils.check_dir_file_exists(output_path, f"{output_root}.mztab")
+    config, model = setup_model(
+        model, config, output_path, output_root_name, False
+    )
+    start_time = time.time()
+    with ModelRunner(
+        config,
+        model,
+        output_path,
+        output_root_name if output_root is not None else None,
+        False,
+    ) as runner:
+        logger.info("Performing database search on:")
+        for peak_file in peak_path:
+            logger.info("  %s", peak_file)
+
+        logger.info("Using the following FASTA file:")
+        logger.info("  %s", fasta_path)
+
+        runner.db_search(
+            peak_path,
+            fasta_path,
+            str((output_path / output_root).with_suffix(".mztab")),
+        )
+        utils.log_run_report(start_time=start_time, end_time=time.time())
+
+
 @main.command(cls=_SharedParams)
 @click.argument(
     "train_peak_path",

casanovo/config.py

@@ -18,6 +18,7 @@
 _config_deprecated = dict(
     every_n_train_steps="val_check_interval",
     max_iters="cosine_schedule_period_iters",
+    max_length="max_peptide_len",
     save_top_k=None,
     model_save_folder_path=None,
 )
@@ -61,7 +62,7 @@ class Config:
         n_layers=int,
         dropout=float,
         dim_intensity=int,
-        max_length=int,
+        max_peptide_len=int,
         residues=dict,
         n_log=int,
         tb_summarywriter=bool,

casanovo/config.yaml

@@ -5,20 +5,22 @@
 
 ###
 # The following parameters can be modified when running inference or when
-# fine-tuning an existing Casanovo model.
+# fine-tuning an existing Casanovo model. They also affect database search
+# parameters when running Casanovo in DB-search mode.
 ###
 
 # Max absolute difference allowed with respect to observed precursor m/z.
-# Predictions outside the tolerance range are assigned a negative peptide score.
+# denovo: Predictions outside the tolerance range are assigned a negative peptide score.
+# db-search: Select candidate peptides within the specified precursor m/z tolerance.
 precursor_mass_tol: 50  # ppm
 # Isotopes to consider when comparing predicted and observed precursor m/z's.
 isotope_error_range: [0, 1]
-# The minimum length of predicted peptides.
+# The minimum length of considered peptides.
 min_peptide_len: 6
+# The maximum length of considered peptides.
+max_peptide_len: 100
 # Number of spectra in one inference batch.
 predict_batch_size: 1024
-# Number of beams used in beam search.
-n_beams: 1
 # Number of PSMs for each spectrum.
 top_match: 1
 # The hardware accelerator to use. Must be one of:
@@ -29,6 +31,42 @@ accelerator: "auto"
 # number will be automatically selected for based on the chosen accelerator.
 devices:
 
+
+###
+# The following parameters are unique to Casanovo's inference/finetuning mode.
+###
+
+# Number of beams used in beam search.
+n_beams: 1
+
+
+###
+# The following parameters are unique to Casanovo's database search mode.
+###
+
+# Enzyme for in silico digestion, used to generate candidate peptides.
+# See pyteomics.parser.expasy_rules for valid enzymes.
+# Can also take a regex to specify custom digestion rules.
+enzyme: "trypsin"
+# Digestion type for candidate peptide generation.
+# full: standard digestion. 
+# semi: Include products of semi-specific cleavage.
+# non-specific: Include products of non-specific cleavage.
+digestion: "full"
+# Number of allowed missed cleavages when digesting protein.
+missed_cleavages: 0
+# Maximum number of variable amino acid modifications per peptide,
+# None generates all possible isoforms as candidates.
+max_mods: 1
+# Select which modifications from the vocabulary can be used in candidate creation.
+# Format: Comma-separated list of "aa:mod_residue", 
+# where aa is a standard amino acid (or "nterm" for an N-terminal mod)
+# and mod_residue is a key from the "residues" dictionary.
+# Example: "M:M+15.995,nterm:+43.006"
+allowed_fixed_mods: "C:C+57.021"
+allowed_var_mods: "M:M+15.995,N:N+0.984,Q:Q+0.984,nterm:+42.011,nterm:+43.006,nterm:-17.027,nterm:+43.006-17.027"
+
+
 ###
 # The following parameters should only be modified if you are training a new
 # Casanovo model from scratch.
@@ -77,8 +115,6 @@ dropout: 0.0
 # Number of dimensions to use for encoding peak intensity.
 # Projected up to `dim_model` by default and summed with the peak m/z encoding.
 dim_intensity:
-# Max decoded peptide length.
-max_length: 100
 # The number of iterations for the linear warm-up of the learning rate.
 warmup_iters: 100_000
 # The number of iterations for the cosine half period of the learning rate.