OpenMS
SequestInfile.h
Go to the documentation of this file.
1 // Copyright (c) 2002-2023, The OpenMS Team -- EKU Tuebingen, ETH Zurich, and FU Berlin
2 // SPDX-License-Identifier: BSD-3-Clause
3 //
4 // --------------------------------------------------------------------------
5 // $Maintainer: Timo Sachsenberg $
6 // $Authors: Martin Langwisch $
7 // --------------------------------------------------------------------------
8 
9 #pragma once
10 
12 
13 #include <map>
14 
15 
16 namespace OpenMS
17 {
25  class OPENMS_DLLAPI SequestInfile
26  {
27 public:
30 
32  SequestInfile(const SequestInfile & sequest_infile);
33 
35  virtual ~SequestInfile();
36 
38  SequestInfile & operator=(const SequestInfile & sequest_infile);
39 
41  bool operator==(const SequestInfile & sequest_infile) const;
42 
47  void store(const String & filename);
48 
51 
53  const String & getDatabase() const;
55  void setDatabase(const String & database);
56 
60  void setNeutralLossesForIons(const String & neutral_losses_for_ions);
61 
63  const String & getIonSeriesWeights() const;
65  void setIonSeriesWeights(const String & ion_series_weights);
66 
68  const String & getPartialSequence() const;
70  void setPartialSequence(const String & partial_sequence);
71 
75  void setSequenceHeaderFilter(const String & sequence_header_filter);
76 
78  const String & getProteinMassFilter() const;
80  void setProteinMassFilter(const String & protein_mass_filter);
81 
82 
84  float getPeakMassTolerance() const;
86  void setPeakMassTolerance(float peak_mass_tolerance);
87 
91  void setPrecursorMassTolerance(float precursor_mass_tolerance);
92 
94  float getMatchPeakTolerance() const;
96  void setMatchPeakTolerance(float match_peak_tolerance);
97 
99  float getIonCutoffPercentage() const;
101  void setIonCutoffPercentage(float ion_cutoff_percentage);
102 
106  void setPeptideMassUnit(Size peptide_mass_unit);
107 
111  void setOutputLines(Size output_lines);
112 
118  Size setEnzyme(const String& enzyme_name);
119 
123  void setMaxAAPerModPerPeptide(Size max_aa_per_mod_per_peptide);
124 
128  void setMaxModsPerPeptide(Size max_mods_per_peptide);
129 
143  void setNucleotideReadingFrame(Size nucleotide_reading_frame);
144 
148  void setMaxInternalCleavageSites(Size max_internal_cleavage_sites);
149 
153  void setMatchPeakCount(Size match_peak_count);
154 
158  void setMatchPeakAllowedError(Size match_peak_allowed_error);
159 
160 
162  bool getShowFragmentIons() const;
164  void setShowFragmentIons(bool show_fragments);
165 
169  void setPrintDuplicateReferences(bool print_duplicate_references);
170 
174  void setRemovePrecursorNearPeaks(bool remove_precursor_near_peaks);
175 
177  bool getMassTypeParent() const;
179  void setMassTypeParent(bool mass_type_parent);
180 
182  bool getMassTypeFragment() const;
184  void setMassTypeFragment(bool mass_type_fragment);
185 
187  bool getNormalizeXcorr() const;
189  void setNormalizeXcorr(bool normalize_xcorr);
190 
194  void setResiduesInUpperCase(bool residues_in_upper_case);
195 
199  void addEnzymeInfo(std::vector<String> & enzyme_info);
200 
202  const std::map<String, std::vector<String> > & getModifications() const;
203 
215  void handlePTMs(const String & modification_line, const String & modifications_filename, const bool monoisotopic);
216 
217 protected:
219  const std::map<String, std::vector<String> > & getEnzymeInfo_() const;
220 
223 
224  std::map<String, std::vector<String> > enzyme_info_;
232 
237 
257 
258 
266 
267  std::map<String, std::vector<String> > PTMname_residues_mass_type_;
268  };
269 
270 } // namespace OpenMS
271 
Sequest input file adapter.
Definition: SequestInfile.h:26
void setShowFragmentIons(bool show_fragments)
sets whether fragment ions shall be displayed
String protein_mass_filter_
Definition: SequestInfile.h:231
void setIonSeriesWeights(const String &ion_series_weights)
sets the weights for the a-, b-, c-, d-, v-, w-, x-, y- and z-ion series
Size getMaxInternalCleavageSites() const
returns the maximum number of internal cleavage sites
void setOutputLines(Size output_lines)
sets the number of peptides to be displayed
void handlePTMs(const String &modification_line, const String &modifications_filename, const bool monoisotopic)
bool getResiduesInUpperCase() const
returns whether residues are in upper case
void setMatchPeakTolerance(float match_peak_tolerance)
sets the match peak tolerance
float match_peak_tolerance_
minimum distance between two experimental peaks
Definition: SequestInfile.h:235
void setMatchPeakAllowedError(Size match_peak_allowed_error)
sets the number of top abundant peaks that are allowed not to match with a theoretical peak
void setMaxAAPerModPerPeptide(Size max_aa_per_mod_per_peptide)
sets the maximum number of amino acids containing the same modification in a peptide
Size max_mods_per_peptide_
maximum number of modifications per peptide
Definition: SequestInfile.h:242
const String & getPartialSequence() const
returns the partial sequences (space delimited) that have to occur in the theoretical spectra
const String & getDatabase() const
returns the used database
SequestInfile & operator=(const SequestInfile &sequest_infile)
assignment operator
Size getMaxModsPerPeptide() const
returns the maximum number of modifications that are allowed in a peptide
Size output_lines_
number of peptides to be displayed
Definition: SequestInfile.h:239
void setPeakMassTolerance(float peak_mass_tolerance)
sets the peak mass tolerance
void setIonCutoffPercentage(float ion_cutoff_percentage)
sets the ion cutoff of the ratio matching theoretical peaks/theoretical peaks
void addEnzymeInfo(std::vector< String > &enzyme_info)
void setNucleotideReadingFrame(Size nucleotide_reading_frame)
void setResiduesInUpperCase(bool residues_in_upper_case)
sets whether residues are in upper case
void setPrintDuplicateReferences(bool print_duplicate_references)
sets whether all proteins containing a found peptide should be displayed
void setMaxModsPerPeptide(Size max_mods_per_peptide)
set the maximum number of modifications that are allowed in a peptide
const String & getNeutralLossesForIons() const
returns whether neutral losses are considered for the a-, b- and y-ions
float ion_cutoff_percentage_
cutoff of the ratio matching theoretical peaks/theoretical peaks
Definition: SequestInfile.h:236
const std::map< String, std::vector< String > > & getModifications() const
return the modifications (the modification names map to the affected residues, the mass change and th...
void setProteinMassFilter(const String &protein_mass_filter)
sets the protein mass filter (either min and max mass, or mass and tolerance value in percent)
String ion_series_weights_
weights for the a-; b-; c-; d-; v-; w-; x-; y- and z-ion series; space delimited
Definition: SequestInfile.h:228
float getPrecursorMassTolerance() const
returns the precursor mass tolerance
float precursor_mass_tolerance_
tolerance for matching a theoretical to an experimental peptide
Definition: SequestInfile.h:233
Size getEnzymeNumber() const
returns the enzyme used for cleavage (by means of the number from a list of enzymes)
void store(const String &filename)
bool getRemovePrecursorNearPeaks() const
return whether peaks near (15 amu) the precursor peak are removed
void setRemovePrecursorNearPeaks(bool remove_precursor_near_peaks)
sets whether peaks near (15 amu) the precursor peak are removed
bool residues_in_upper_case_
whether residues are in upper case
Definition: SequestInfile.h:265
const std::map< String, std::vector< String > > & getEnzymeInfo_() const
returns the enzyme list
Size max_internal_cleavage_sites_
maximum number of internal cleavage sites
Definition: SequestInfile.h:254
float getIonCutoffPercentage() const
returns the the cutoff of the ratio matching theoretical peaks/theoretical peaks
const String & getProteinMassFilter() const
returns the protein mass filter (either min and max mass, or mass and tolerance value in percent)
Size nucleotide_reading_frame_
Definition: SequestInfile.h:243
float peak_mass_tolerance_
tolerance for matching a theoretical to an experimental peak
Definition: SequestInfile.h:234
Size getPeptideMassUnit() const
returns the peptide mass unit
String neutral_losses_for_ions_
whether neutral losses are considered for the a-; b- and y-ions (e.g. 011 for b- and y-ions)
Definition: SequestInfile.h:227
void setSequenceHeaderFilter(const String &sequence_header_filter)
sets the sequences (space delimited) that have to occur, or be absent (preceded by a tilde) in the he...
Size getMatchPeakCount() const
returns the number of top abundant peaks to match with theoretical ones
float getMatchPeakTolerance() const
returns the match peak tolerance
void setMassTypeParent(bool mass_type_parent)
sets the mass type of the parent (0 - monoisotopic, 1 - average mass)
bool getNormalizeXcorr() const
returns whether normalized xcorr values are displayed
Size peptide_mass_unit_
peptide mass unit (0 = amu; 1 = mmu; 2 = ppm)
Definition: SequestInfile.h:238
bool remove_precursor_near_peaks_
whether peaks near (15 amu) the precursor peak are removed
Definition: SequestInfile.h:261
String partial_sequence_
space-delimited list of sequence parts that have to occur in the theoretical spectra
Definition: SequestInfile.h:229
bool show_fragment_ions_
whether to display fragment ions
Definition: SequestInfile.h:259
void setMatchPeakCount(Size match_peak_count)
sets the number of top abundant peaks to with theoretical ones
std::map< String, std::vector< String > > enzyme_info_
an endline-delimited list of enzymes; each with cutting direction 0 (N to C) /1; cuts after (list of ...
Definition: SequestInfile.h:224
Size getMaxAAPerModPerPeptide() const
returns the maximum number of amino acids containing the same modification in a peptide
String sequence_header_filter_
space-delimited list of sequences that have to occur or be absent (preceded by a tilde) in a protein ...
Definition: SequestInfile.h:230
Size match_peak_count_
number of the top abundant peaks to match with theoretical one
Definition: SequestInfile.h:255
SequestInfile()
default constructor
bool operator==(const SequestInfile &sequest_infile) const
equality operator
String snd_database_
second database used
Definition: SequestInfile.h:226
void setNormalizeXcorr(bool normalize_xcorr)
sets whether normalized xcorr values are displayed
bool mass_type_parent_
mass type of the parent peak (0 - monoisotopic; 1 - average)
Definition: SequestInfile.h:262
void setStandardEnzymeInfo_()
returns some standard enzymes (used to initialize the enzyme list)
String database_
database used
Definition: SequestInfile.h:225
String getEnzymeName() const
returns the enzyme used for cleavage
Size enzyme_number_
number of the enzyme used for cleavage
Definition: SequestInfile.h:240
bool print_duplicate_references_
whether all proteins containing a found peptide should be displayed
Definition: SequestInfile.h:260
const String & getIonSeriesWeights() const
returns the weights for the a-, b-, c-, d-, v-, w-, x-, y- and z-ion series
Size getNucleotideReadingFrame() const
returns the nucleotide reading frame
void setPrecursorMassTolerance(float precursor_mass_tolerance)
sets the precursor mass tolerance
Size getMatchPeakAllowedError() const
returns the number of top abundant peaks that are allowed not to match with a theoretical peak
std::map< String, std::vector< String > > PTMname_residues_mass_type_
the modification names map to the affected residues, the mass change and the type
Definition: SequestInfile.h:267
void setPartialSequence(const String &partial_sequence)
sets the partial sequences (space delimited) that have to occur in the theoretical spectra
virtual ~SequestInfile()
destructor
bool mass_type_fragment_
mass type of fragment peaks (0 - monoisotopic; 1 - average)
Definition: SequestInfile.h:263
bool getShowFragmentIons() const
returns whether fragment ions shall be displayed
void setDatabase(const String &database)
sets the used database
Size match_peak_allowed_error_
number of peaks that may lack this test
Definition: SequestInfile.h:256
void setMassTypeFragment(bool mass_type_fragment)
sets the mass type of the fragments (0 - monoisotopic, 1 - average mass)
Size setEnzyme(const String &enzyme_name)
sets the enzyme used for cleavage (by means of the number from a list of enzymes)
bool getMassTypeFragment() const
return the mass type of the fragments (0 - monoisotopic, 1 - average mass)
float getPeakMassTolerance() const
returns the peak mass tolerance
void setNeutralLossesForIons(const String &neutral_losses_for_ions)
sets whether neutral losses are considered for the a-, b- and y-ions
bool getPrintDuplicateReferences() const
returns whether all proteins containing a found peptide should be displayed
bool normalize_xcorr_
whether to display normalized xcorr values
Definition: SequestInfile.h:264
const String & getSequenceHeaderFilter() const
returns the sequences (space delimited) that have to occur, or be absent (preceded by a tilde) in the...
void setPeptideMassUnit(Size peptide_mass_unit)
sets the peptide mass unit
Size max_AA_per_mod_per_peptide_
maximum number of amino acids containing the same modification in a peptide
Definition: SequestInfile.h:241
SequestInfile(const SequestInfile &sequest_infile)
copy constructor
const String getEnzymeInfoAsString() const
returns the enzyme list as a string
bool getMassTypeParent() const
return the mass type of the parent (0 - monoisotopic, 1 - average mass)
Size getOutputLines() const
return the number of peptides to be displayed
void setMaxInternalCleavageSites(Size max_internal_cleavage_sites)
sets the maximum number of internal cleavage sites
A more convenient string class.
Definition: String.h:34
size_t Size
Size type e.g. used as variable which can hold result of size()
Definition: Types.h:101
Main OpenMS namespace.
Definition: FeatureDeconvolution.h:22