Fragment Index Construction

The database section of the configuration file determines how the internal fragment index data structure is generated.

{
    "bucket_size": 8192,
    "enzyme": {
        "missed_cleavages": 2,
        "min_len": 7,
        "max_len": 50,
        "cleave_at": "KR",
        "restrict": "P",
        "c_terminal": true,
        "semi_enzymatic": false
    },
    "peptide_min_mass": 500.0,
    "peptide_max_mass": 5000.0,
    "ion_kinds": [
        "b",
        "y"
    ],
    "min_ion_index": 2,
    "max_variable_mods": 2,
    "static_mods": {
        "C": 57.0214
    },
    "variable_mods": {
        "M": 15.9949
    },
    "decoy_tag": "rev_",
    "generate_decoys": true,
    "fasta": "s3://sage-benchmarking/fasta/human_contam.fasta"
}

Bucket size

This parameter can be used to tune performance of Sage. This value sets the number of fragment ions within each "bucket" in the internal index datastructure. This value will always be set to the next largest power of 2.

A smaller number (8192 is the minimum) is suitable for high resolution MS/MS spectra, since not many buckets will need to be searched. Low resolution MS/MS spectra will need to search more buckets, so increasing the size of the bucket will lower the total number of internal buckets. A good starting point is to use 65536 for ion-trap data, but the optimal value for your search parameters and files might require empirical tuning.

Enzyme

The enzyme section contains parameters related to the enzyme used for digestion. The default enzyme is trypsin, with the parameters specified below.

• missed_cleavages: Integer. The number of missed cleavages for tryptic digest (default: 1).

• min_len: Integer. The minimum amino acid (AA) length of peptides to search (default: 5).

• max_len: Integer. The maximum AA length of peptides to search (default: 50).

• cleave_at: String. Amino acids to cleave at (default: 'KR').

  The cleave_at parameter can also be used to specify alternative digestion schemes:

  • Non-enzymatic: cleave_at = "" - All potential peptides between min_len and max_len will be generated from the sequence
  • No digestion: cleave_at = "$" - FASTA entries will be used as-is, subject to min_len and max_len options

• restrict: Single character string. Do not cleave if this amino acid follows the cleavage site (default: 'P').

• c_terminal: Boolean. Cleave at the C-terminus of matching amino acids (default:true).

• semi_enzymatic: Boolean. Perform a semi-enzymatic digest (default:false).

Fragment Settings

• peptide_min_mass: Float. The minimum monoisotopic mass of peptides to fragment in silico (default: 500.0).
• peptide_max_mass: Float. The maximum monoisotopic mass of peptides to fragment in silico (default: 5000.0).
• ion_kinds: List of strings. Which fragment ions to produce? Allowed values: "a", "b", "c", "x", "y", "z". (default: ["b", "y"])
• min_ion_index: Integer. Do not generate b1..bN or y1..yN ions for preliminary searching if min_ion_index = N. Does not affect full scoring of PSMs (default: 2).

Modifications

Static Modifications

• static_mods Dictionary with characters as keys and floats as values. Represents static modifications applied to amino acids or termini (default: ). Static modifications are applied after variable modifications

Example: Apply a static modification of 304.207 to the N-terminus of the peptide and lysine, and 57.0215 to cysteine.

{
    "static_mods": {
        "^": 304.207,
        "K": 304.207,
        "C": 57.0215
    }
}

Variable Modifications

• max_variable_mods: Integer. Limit k-combinations of variable modifications (default: 2).
• variable_mods: Dictionary with characters as keys and list of floats (or single floats) as values. Represents variable modifications applied to amino acids or termini (default: ).

Example: Apply a variable modification of 15.9949 to methionine, 49.2022 to the C-terminus of the peptide, 42.0 to the N-terminus of the protein, and 111.0 to the C-terminus of the protein, in addition to pyro-glutamine/pyro-glutamic acid. Allow only up to 3 variable modifications in total.

{
    "max_variable_mods": 3,
    "variable_mods": {
        "M": [15.9949], 
        "^Q": [-17.026549],
        "^E": [-18.010565],
        "$": [49.2022],
        "[": 42.0,
        "]": 111.0
    }
}

Modification Syntax:

• "^X": Modification to be applied to amino acid X if it appears at the N-terminus of a peptide
• "$X": Modification to be applied to amino acid X if it appears at the C-terminus of a peptide
• "[X": Modification to be applied to amino acid X if it appears at the N-terminus of a protein
• "]X": Modification to be applied to amino acid X if it appears at the C-terminus of a protein

Fasta Database and Decoy Generation

• decoy_tag: String. The tag used to identify decoy entries in the FASTA database (default: "rev_").
• generate_decoys: Boolean. If true, ignore decoys in the FASTA database matching decoy_tag, and generate internally reversed peptides (default: false).
• fasta: String. The path to the FASTA file, either a local path or s3 object URI.

Target-decoy competition is key to controlling the false discovery rate in proteomics experiments. Sage can use decoy sequences included in the supplied FASTA file, or it can generate internal sequences (recommended). Sage reverses tryptic peptides (not proteins), so that the picked-peptide (opens in a new tab) approach to FDR can be used.

If generate_decoys is set to true (or unspecified), then decoy sequences in the FASTA database matching decoy_tag will be ignored, and Sage will internally generate decoys.

Internally generated decoys will have protein accessions matching "{decoy_tag}{accession}", e.g. if decoy_tag is "rev_" then a protein accession like "rev_sp|P01234|HUMAN" will be listed in the output file.