1- # ' Filtering msfragger Data
1+ # ' Filtering MSFragger Data
2+ # '
3+ # ' Filtering MSFragger data. In this implementation, the peptide-level filter
4+ # ' optimizes both ppm and one of Expectation or PeptideProphet Probability
5+ # ' thresholds to achieve maximum number of peptide identifications within a
6+ # ' given FDR constraint.
27# '
3- # ' Filtering msfragger data. In this implementation, the peptide-level filter
4- # ' optimizes both ppm and one of Expectation or PeptideProphet Probability thresholds
5- # ' to achieve maximum number of peptide identifications within a given FDR constraint.
68# ' The accession-level filter optimizes based on `peptides_per_1000aa`, so
79# ' \code{\link{compute_num_peptides_per_1000aa}} must be used first.
810# '
911# ' @md
1012# '
11- # ' @param msnid (MSnID object) collated msfragger output
13+ # ' @param msnid (MSnID object) collated MSFragger output
1214# ' @param fdr.max (numeric) Maximum acceptable FDR. Default is 0.01 (1%).
1315# ' @param level (character) Level at which to perform FDR filter. The name of a
1416# ' column in `psms(msnid)`. Currently, only `"peptide"` or `"accession"` are
15- # ' supported. The added level `SiteID` makes sense only for PTM data and
17+ # ' supported. The added level `" SiteID" ` makes sense only for PTM data and
1618# ' first requires mapping of the modification site using
17- # ' `MSnID::map_mod_sites`.
18- # ' @param filtering_criterion (character) One of "evalue" which is
19- # ' expectation value or "pp_prob" - peptide prophet probability. Default is
20- # ' "pp_prob".
19+ # ' `MSnID::map_mod_sites`.
20+ # ' @param filtering_criterion (character) One of `"evalue"` which is expectation
21+ # ' value or `"pp_prob"` - peptide prophet probability. Default is "pp_prob".
2122# ' @param n.iter.grid (numeric) number of grid-distributed evaluation points.
2223# ' @param n.iter.nm (numeric) number of iterations for Nelder-Mead optimization
2324# ' algorithm.
24- # ' @param ... arguments passed to `filter_msfragger_data`.
2525# '
26- # ' @return (MSnID object) filtered msfragger output
26+ # ' @return (MSnID object) filtered MSFragger output
2727# '
28- # ' @seealso
29- # ' \code{\link[MSnID]{MSnIDFilter}}
30- # ' \code{\link[MSnID]{optimize_filter}}
31- # ' \code{\link[MSnID]{apply_filter}}
28+ # ' @seealso \code{\link[MSnID]{MSnIDFilter}}
29+ # ' \code{\link[MSnID]{optimize_filter}} \code{\link[MSnID]{apply_filter}}
3230# '
33- # ' @importFrom MSnID MSnIDFilter optimize_filter
34- # ' mass_measurement_error apply_filter
31+ # ' @importFrom MSnID MSnIDFilter optimize_filter mass_measurement_error
32+ # ' apply_filter
3533
3634
3735# ' @export
3836filter_msfragger_data <- function (msnid ,
39- level ,
40- filtering_criterion = c(" pp_prob" " evalue"
41- fdr.max = 0.01 ,
42- n.iter.grid = 500 ,
43- n.iter.nm = 100 ){
37+ level ,
38+ filtering_criterion = c(" pp_prob" " evalue"
39+ fdr.max = 0.01 ,
40+ n.iter.grid = 500 ,
41+ n.iter.nm = 100 ){
4442
4543 # Clean up on exit
4644 on.exit(rm(list = ls()))
@@ -49,18 +47,18 @@ filter_msfragger_data <- function(msnid,
4947 # Check input
5048 level <- match.arg(level , choices = c(" peptide" " accession" " SiteID"
5149 filtering_criterion <- match.arg(filtering_criterion )
52-
53- if (level == " SiteID" & ! (" SiteID" %in% names(msnid )))
54- stop(" Column 'SiteID' is not in the MSnID object. Please map the PTMs first."
50+
51+ if (level == " SiteID" & ! (" SiteID" %in% names(msnid ))) {
52+ stop(" Column 'SiteID' is not in the MSnID object. Please map the PTMs first."
53+ }
5554
5655 keep_cols <- c(level , " isDecoy" # columns to calculate FDR
5756
5857 # Create MSnID of minimum size
5958 suppressMessages(msnid_small <- MSnID())
60-
59+
6160 # Setup
6261 if (level == " accession"
63-
6462 # Add filter criteria column
6563 keep_cols <- c(keep_cols , " peptides_per_1000aa"
6664 msnid_small @ psms <- unique(msnid @ psms [, keep_cols , with = FALSE ])
@@ -69,36 +67,41 @@ filter_msfragger_data <- function(msnid,
6967 filtObj <- MSnIDFilter(msnid_small )
7068 filtObj $ peptides_per_1000aa <- list (comparison = " >" threshold = 1 )
7169 method <- " SANN"
70+
7271 } else {
73- # Choose filter object probability value
74- if (filtering_criterion == " evalue"
75- msnid $ msmsScore <- - log10(msnid $ Expectation )
76- }
77- if (filtering_criterion == " pp_prob"
78- msnid $ msmsScore <- msnid $ `PeptideProphet Probability`
79- }
80- # Create columns for peptide filtering
81- # Can not use data.table syntax if the msnid has been modified at all,
82- # as it results in the "Invalid .internal.selfref" warning and
83- # columns not being created.
84- msnid $ absParentMassErrorPPM <- abs(mass_measurement_error(msnid ))
85-
86- # Add filter criteria columns
87- keep_cols <- c(keep_cols , " msmsScore" " absParentMassErrorPPM"
88- msnid_small @ psms <- unique(msnid @ psms [, keep_cols , with = FALSE ])
89-
90- # Create filter object
91- filtObj <- MSnIDFilter(msnid_small )
92- filtObj $ absParentMassErrorPPM <- list (comparison = " <" threshold = 10 )
93- if (filtering_criterion == " evalue"
94- filtObj $ msmsScore <- list (comparison = " >" threshold = 2 )
95- }
96- if (filtering_criterion == " pp_prob"
97- filtObj $ msmsScore <- list (comparison = " >" threshold = 0.99 )
98- }
99- method <- " Nelder-Mead"
72+ # Choose filter object probability value
73+ if (filtering_criterion == " evalue"
74+ msnid $ msmsScore <- - log10(msnid $ Expectation )
75+ }
76+
77+ if (filtering_criterion == " pp_prob"
78+ msnid $ msmsScore <- msnid $ `PeptideProphet Probability`
79+ }
80+
81+ # Create columns for peptide filtering
82+ # Can not use data.table syntax if the msnid has been modified at all,
83+ # as it results in the "Invalid .internal.selfref" warning and
84+ # columns not being created.
85+ msnid $ absParentMassErrorPPM <- abs(mass_measurement_error(msnid ))
86+
87+ # Add filter criteria columns
88+ keep_cols <- c(keep_cols , " msmsScore" " absParentMassErrorPPM"
89+ msnid_small @ psms <- unique(msnid @ psms [, keep_cols , with = FALSE ])
90+
91+ # Create filter object
92+ filtObj <- MSnIDFilter(msnid_small )
93+ filtObj $ absParentMassErrorPPM <- list (comparison = " <" threshold = 10 )
94+
95+ if (filtering_criterion == " evalue"
96+ filtObj $ msmsScore <- list (comparison = " >" threshold = 2 )
97+ }
98+
99+ if (filtering_criterion == " pp_prob"
100+ filtObj $ msmsScore <- list (comparison = " >" threshold = 0.99 )
101+ }
102+
103+ method <- " Nelder-Mead"
100104 }
101-
102105
103106 # step 1
104107 filtObj.grid <- optimize_filter(filtObj ,
@@ -107,12 +110,17 @@ filter_msfragger_data <- function(msnid,
107110 method = " Grid"
108111 level = level ,
109112 n.iter = n.iter.grid )
113+
110114 # step 2
111115 filtObj.nm <- optimize_filter(filtObj.grid ,
112116 msnid_small ,
113117 fdr.max = fdr.max ,
114118 method = method ,
115119 level = level ,
116120 n.iter = n.iter.nm )
117- return (apply_filter(msnid , filtObj.nm ))
121+
122+ msnid <- apply_filter(msnid , filtObj.nm )
123+
124+ return (msnid )
118125}
126+
0 commit comments