Package {iq}

Type:	Package
Title:	Protein Quantification in Mass Spectrometry-Based Proteomics
Version:	2.2.0
Maintainer:	Thang V. Pham <t.pham@amsterdamumc.nl>
Description:	An implementation of the MaxLFQ algorithm by Cox et al. (2014) <doi:10.1074/mcp.M113.031591> using least-squares analysis in a comprehensive pipeline for processing proteomics data in data-independent acquisition mode (Pham et al. 2020 <doi:10.1093/bioinformatics/btz961>, and a new algorithm using the conjugate gradient method Pham et al. 2026 <doi:10.1021/acs.jproteome.5c01038>). It offers additional options for protein quantification using the N most intense fragment ions, using all fragment ions, the median polish algorithm by Tukey (1977, ISBN:0201076160), a robust linear model, and fragment weighting. A new iq format enables the processing of large-scale datasets. In general, the tool can be used to integrate multiple proportional observations into a single quantitative value.
Depends:	R (≥ 4.0.0)
License:	BSD_3_clause + file LICENSE
Encoding:	UTF-8
LazyData:	true
Suggests:	knitr, rmarkdown, arrow
VignetteBuilder:	knitr
URL:	https://github.com/tvpham/iq
BugReports:	https://github.com/tvpham/iq/issues
NeedsCompilation:	yes
Packaged:	2026-11-06 02:36:09 UTC; p032181
Author:	Thang V. Pham [aut, cre, cph], Alex A. Henneman [ctb]
Repository:	CRAN
Date/Publication:	2026-11-06 12:00:02 UTC

Group connected samples

Description

Two samples are directly connected if they share at least one quantified fragment ion. Two samples in a connected component are either directly connected or indirectly via other samples.

Usage

connected_component(X)

Arguments

Value

A vector with length equal to the number of columns of the input containing membership of the connected components.

Author(s)

Thang V. Pham

References

Pham TV, Tran CTM, Henneman AA, Pham LHC, Le DG, Can AH, Bui PHL, Piersma SR, Jimenez CR. Boosting the speed and accuracy of protein quantification algorithms in mass spectrometry-based proteomics. Journal of Proteome Research 2026 Feb 6; 25(2):1198-1203.

Creating a list of matrices of fragment ion intensities for all proteins

Description

For each protein, a numerical matrix is formed where the columns are samples and rows are fragment ions.

Usage

create_protein_list(preprocessed_data)

Arguments

Value

A list where each element contains the quantitative data of a protein. The column names are sample names and the row names fragment ions.

Author(s)

Thang V. Pham

References

Pham TV, Tran CTM, Henneman AA, Pham LHC, Le DG, Can AH, Bui PHL, Piersma SR, Jimenez CR. Boosting the speed and accuracy of protein quantification algorithms in mass spectrometry-based proteomics. Journal of Proteome Research 2026 Feb 6; 25(2):1198-1203.

Pham TV, Henneman AA, Jimenez CR. iq: an R package to estimate relative protein abundances from ion quantification in DIA-MS-based proteomics. Bioinformatics 2020 Apr 15;36(8):2611-2613.

See Also

preprocess

Examples

data("spikeins")
head(spikeins)
# This example set of spike-in proteins has been 'median-normalized'.
norm_data <- iq::preprocess(spikeins, median_normalization = FALSE, pdf_out = NULL)
protein_list <- iq::create_protein_list(norm_data)

Protein quantification for a list of proteins

Description

Travels through the input list and quantifies all proteins one by one.

Usage

create_protein_table(protein_list, method = "maxLFQ", ...)

Arguments

Value

A list of two components is returned

For an empty protein_list, estimate is a matrix without rows and annotation is empty.

Author(s)

Thang V. Pham

References

Pham TV, Tran CTM, Henneman AA, Pham LHC, Le DG, Can AH, Bui PHL, Piersma SR, Jimenez CR. Boosting the speed and accuracy of protein quantification algorithms in mass spectrometry-based proteomics. Journal of Proteome Research 2026 Feb 6; 25(2):1198-1203.

Pham TV, Henneman AA, Jimenez CR. iq: an R package to estimate relative protein abundances from ion quantification in DIA-MS-based proteomics. Bioinformatics 2020 Apr 15;36(8):2611-2613.

See Also

create_protein_list, maxLFQ, median_polish, topN, meanInt

Examples

data("spikeins")
# This example set of spike-in proteins has been 'median-normalized'.
norm_data <- iq::preprocess(spikeins, median_normalization = FALSE, pdf_out = NULL)
protein_list <- iq::create_protein_list(norm_data)
result <- iq::create_protein_table(protein_list)
head(result)

Protein annotation extraction

Description

Extracts annotation columns from a long-format input

Usage

extract_annotation(protein_ids, quant_table, primary_id = "PG.ProteinGroups",
                   annotation_columns = NULL)

Arguments

Value

A table of proteins and associated annotation extracted from the input.

Author(s)

Thang V. Pham

References

Pham TV, Tran CTM, Henneman AA, Pham LHC, Le DG, Can AH, Bui PHL, Piersma SR, Jimenez CR. Boosting the speed and accuracy of protein quantification algorithms in mass spectrometry-based proteomics. Journal of Proteome Research 2026 Feb 6; 25(2):1198-1203.

Pham TV, Henneman AA, Jimenez CR. iq: an R package to estimate relative protein abundances from ion quantification in DIA-MS-based proteomics. Bioinformatics 2020 Apr 15;36(8):2611-2613.

See Also

preprocess

Examples

data("spikeins")
extra_names <- iq::extract_annotation(levels(spikeins$PG.ProteinGroups),
                                      spikeins,
                                      annotation_columns = c("PG.Genes", "PG.ProteinNames"))

The MaxLFQ algorithm

Description

A fast implementation of the MaxLFQ algorithm.

Usage

fast_MaxLFQ(norm_data, row_names = NULL, col_names = NULL, verbose = TRUE)

Arguments

Details

The MaxLFQ problem of each protein is solved with the "maxlfq_bit" method of process_matrix.

Value

A list is returned with two components

Author(s)

Thang V. Pham

References

Pham TV, Tran CTM, Henneman AA, Pham LHC, Le DG, Can AH, Bui PHL, Piersma SR, Jimenez CR. Boosting the speed and accuracy of protein quantification algorithms in mass spectrometry-based proteomics. Journal of Proteome Research 2026 Feb 6; 25(2):1198-1203.

Pham TV, Henneman AA, Jimenez CR. iq: an R package to estimate relative protein abundances from ion quantification in DIA-MS-based proteomics. Bioinformatics 2020 Apr 15;36(8):2611-2613.

See Also

fast_read, fast_preprocess

Data filtering and normalization

Description

Filters out low intensities and performs median normalization.

Usage

fast_preprocess(quant_table,
                median_normalization = TRUE,
                log2_intensity_cutoff = 0,
                pdf_out = "qc-plots-fast.pdf",
                pdf_width = 12,
                pdf_height = 8,
                show_boxplot = TRUE,
                boxplot_max_points = 10000,
                verbose = TRUE)

Arguments

Value

A list is returned with the same components as input data in which low intensities are filtered out and median normalization is performed if requested.

Author(s)

Thang V. Pham

References

Pham TV, Tran CTM, Henneman AA, Pham LHC, Le DG, Can AH, Bui PHL, Piersma SR, Jimenez CR. Boosting the speed and accuracy of protein quantification algorithms in mass spectrometry-based proteomics. Journal of Proteome Research 2026 Feb 6; 25(2):1198-1203.

Pham TV, Henneman AA, Jimenez CR. iq: an R package to estimate relative protein abundances from ion quantification in DIA-MS-based proteomics. Bioinformatics 2020 Apr 15;36(8):2611-2613.

See Also

fast_read

Reading data from an input file

Description

A highly efficient reading of a tab-separated text file for iq processing.

Usage

fast_read(filename,
          sample_id = "R.Condition",
          primary_id = "PG.ProteinGroups",
          secondary_id = c("EG.ModifiedSequence", "FG.Charge", "F.FrgIon", "F.Charge"),
          intensity_col = "F.PeakArea",
          annotation_col = c("PG.Genes", "PG.ProteinNames"),
          filter_string_equal = c("F.ExcludedFromQuantification" = "False"),
          filter_string_not_equal = NULL,
          filter_double_less = c("PG.Qvalue" = "0.01", "EG.Qvalue" = "0.01"),
          filter_double_greater = NULL,
          intensity_col_sep = NULL,
          intensity_col_id = NULL,
          na_string = "0",
          verbose = TRUE)

Arguments

Details

When there are multiple quantitative values per row (a single intensity column split by intensity_col_sep, or several intensity columns), this function will replicate entries in other columns and the secondary_id will be appended with corresponding entries in intensity_col_id when it is provided. Otherwise, integer values 1, 2, 3, etc... will be used.

Value

A list is returned with following components

If the file cannot be read (e.g. a missing file or column) or no values are left after filtering, fast_read stops with an error.

Author(s)

Thang V. Pham

References

Pham TV, Tran CTM, Henneman AA, Pham LHC, Le DG, Can AH, Bui PHL, Piersma SR, Jimenez CR. Boosting the speed and accuracy of protein quantification algorithms in mass spectrometry-based proteomics. Journal of Proteome Research 2026 Feb 6; 25(2):1198-1203.

Pham TV, Henneman AA, Jimenez CR. iq: an R package to estimate relative protein abundances from ion quantification in DIA-MS-based proteomics. Bioinformatics 2020 Apr 15;36(8):2611-2613.

Reading data from a data frame

Description

Reads a long-format data frame for iq processing, with the same filtering rules as fast_read and fast_read_parquet.

Usage

fast_read_data_frame(input_data,
                     sample_id = "Run",
                     primary_id = "Protein.Group",
                     secondary_id = "Precursor.Id",
                     intensity_col = "Precursor.Normalised",
                     annotation_col = NULL,
                     filter_string_equal = NULL,
                     filter_string_not_equal = NULL,
                     filter_double_less = c("Q.Value" = "0.01", "PG.Q.Value" = "0.01"),
                     filter_double_greater = NULL,
                     intensity_col_sep = NULL,
                     intensity_col_id = NULL,
                     na_string = "0",
                     verbose = TRUE)

Arguments

Details

A filter value is missing if it is NA, NaN, empty, or text that is not a number; such rows are kept.

Columns can be character, factor, integer, double or logical. Numbers used as identifiers (e.g. a charge column in secondary_id) are converted to text.

When there are multiple quantitative values per row, the secondary_id is appended with the corresponding entries of intensity_col_id when it is provided. Otherwise, integer values 1, 2, 3, etc... are used.

Value

A list as returned by fast_read.

Author(s)

Thang V. Pham

References

Pham TV, Tran CTM, Henneman AA, Pham LHC, Le DG, Can AH, Bui PHL, Piersma SR, Jimenez CR. Boosting the speed and accuracy of protein quantification algorithms in mass spectrometry-based proteomics. Journal of Proteome Research 2026 Feb 6; 25(2):1198-1203.

Reading data from a parquet file

Description

Reads a long-format parquet file, for example a DIA-NN report.parquet, one row group at a time for iq processing. Only the needed columns are read, and only one row group is held in memory at a time. Requires the arrow package.

Usage

fast_read_parquet(filename,
                  sample_id = "Run",
                  primary_id = "Protein.Group",
                  secondary_id = "Precursor.Id",
                  intensity_col = "Precursor.Normalised",
                  annotation_col = NULL,
                  filter_string_equal = NULL,
                  filter_string_not_equal = NULL,
                  filter_double_less = c("Q.Value" = "0.01", "PG.Q.Value" = "0.01"),
                  filter_double_greater = NULL,
                  intensity_col_sep = NULL,
                  intensity_col_id = NULL,
                  na_string = "0",
                  verbose = TRUE)

Arguments

Details

Filtering and missing values follow fast_read, and the result is the same as fast_read on the same data written as a tab-separated text file. A filter value is missing if it is NA, NaN, empty, or text that is not a number; such rows are kept.

When there are multiple quantitative values per row, the secondary_id is appended with the corresponding entries of intensity_col_id when it is provided. Otherwise, integer values 1, 2, 3, etc... are used.

Value

A list as returned by fast_read.

Author(s)

Thang V. Pham

References

Pham TV, Tran CTM, Henneman AA, Pham LHC, Le DG, Can AH, Bui PHL, Piersma SR, Jimenez CR. Boosting the speed and accuracy of protein quantification algorithms in mass spectrometry-based proteomics. Journal of Proteome Research 2026 Feb 6; 25(2):1198-1203.

Calculate Huber weights for an individual matrix

Description

Calculate Huber weights for an individual matrix

Usage

huber_weight(X, x, c = 1.4826, k = 1.345)

Arguments

Value

A numeric weight vector of size equal to the number of rows of X.

Author(s)

Thang V. Pham

References

Pham TV, Tran CTM, Henneman AA, Pham LHC, Le DG, Can AH, Bui PHL, Piersma SR, Jimenez CR. Boosting the speed and accuracy of protein quantification algorithms in mass spectrometry-based proteomics. Journal of Proteome Research 2026 Feb 6; 25(2):1198-1203.

Protein data from the iq format

Description

Reads the data of one or more proteins from the iq format into a list of matrices, as returned by create_protein_list, for example to plot a protein with plot_protein.

Usage

iq_format_to_protein_list(iqformat_folder, proteins, normalization = TRUE)

Arguments

Details

Only the files of the requested proteins are read. The intensities of an ion in a sample are summed, as in process_iq_format, so that process_matrix on a matrix of the list gives the same result as the row of that protein in the output of process_iq_format with the same method. A protein without values (it has no protein file, e.g. all its values are below the intensity cut-off) gets a matrix without rows, with a warning. An identifier that is not in _protein.tsv stops with an error.

Value

A list with one element for each protein in proteins, named by its identifier: a matrix of log2 intensities with the ions in rows (named as in _ion.tsv) and all samples in columns (named as in _sample.tsv); NA for a missing value.

Author(s)

Thang V. Pham

References

Pham TV, Tran CTM, Henneman AA, Pham LHC, Le DG, Can AH, Bui PHL, Piersma SR, Jimenez CR. Boosting the speed and accuracy of protein quantification algorithms in mass spectrometry-based proteomics. Journal of Proteome Research 2026 Feb 6; 25(2):1198-1203.

See Also

long_format_to_iq_format, process_iq_format, plot_protein, process_matrix

Long format input data to iq format

Description

This function reads long-format input data (a text file or a parquet file), filters it, and writes the results to the iq format. The input is read and written incrementally, so it is never held in memory as a whole.

Usage

long_format_to_iq_format(input_filename, 
                         output_iqformat_folder,
                         sample_id = "Run",
                         primary_id = "Protein.Group",
                         secondary_id = c("Precursor.Id"),
                         intensity_col = "Precursor.Normalised",
                         annotation_col = NULL,
                         filter_string_equal = NULL,
                         filter_string_not_equal = NULL,
                         filter_double_less = c("Q.Value" = "0.01",
                                                "PG.Q.Value" = "0.01"),
                         filter_double_greater = NULL,
                         intensity_col_sep = NULL,
                         intensity_col_id = NULL, 
                         na_string = "0",
                         normalization = "median",
                         log2_intensity_cutoff = 0,
                         pdf_out = "qc-plots-iq-format.pdf",
                         pdf_width = 12,
                         pdf_height = 8,
                         show_boxplot = TRUE,
                         peptide_extractor = NULL,
                         rfasta = NULL,
                         buffer_mb = 1024,
                         verbose = TRUE)

Arguments

Details

The input is read in chunks: blocks of lines of a text file, or row groups of a parquet file (converted to R in slices of rows). The values of each chunk are buffered (buffer_mb) and appended to the files. Only the lists of proteins, samples and ions are kept in memory; for a parquet file, also one row group at a time, as decoded by arrow.

The defaults are those of process (DIA-NN precursor quantities Precursor.Normalised per Run, filtered on Q.Value and PG.Q.Value below 0.01, log2 intensity cut-off 0), so that the same arguments give the same proteins as process.

The preprocessing is that of fast_preprocess: values that are zero or negative are left out, as are values not above log2_intensity_cutoff (by default 0). The protein files contain the intensities as read; the normalization, a shift of each sample in log2 space, is stored in the normalization column of _sample.tsv and applied by process_iq_format. If peptide_extractor is given, the fragment statistics of process (n_fragments, n_peptides, peptides, XIC and, with rfasta, coverage) are computed while the input is read, from all values read as in process (also those left out by the preprocessing), and stored as further columns of _protein.tsv; process_iq_format writes them with the protein annotation. Results are the same as with process on the same input.

Value

NULL, invisibly. The output is written to the folder output_iqformat_folder: _protein.tsv (proteins and annotation, and the fragment statistics if peptide_extractor is given), _sample.tsv (samples and their normalization), _ion.tsv (ions), and one file <i>.tsv for each protein with values (i its row in _protein.tsv), with the sample index, ion index and intensity of each value.

Author(s)

Thang V. Pham

References

Pham TV, Tran CTM, Henneman AA, Pham LHC, Le DG, Can AH, Bui PHL, Piersma SR, Jimenez CR. Boosting the speed and accuracy of protein quantification algorithms in mass spectrometry-based proteomics. Journal of Proteome Research 2026 Feb 6; 25(2):1198-1203.

See Also

fast_read, fast_read_parquet, fast_preprocess, process_iq_format, process

The MaxLFQ algorithm for protein quantification

Description

Estimates protein abundances by aiming to maintain the fragment intensity ratios between samples.

Usage

maxLFQ(X)

Arguments

Value

A list of two components is returned

Author(s)

Thang V. Pham

References

Pham TV, Tran CTM, Henneman AA, Pham LHC, Le DG, Can AH, Bui PHL, Piersma SR, Jimenez CR. Boosting the speed and accuracy of protein quantification algorithms in mass spectrometry-based proteomics. Journal of Proteome Research 2026 Feb 6; 25(2):1198-1203.

Cox J, Hein MY, Luber CA, et al. Accurate proteome-wide label-free quantification by delayed normalization and maximal peptide ratio extraction, termed MaxLFQ. Mol Cell Proteomics. 2014;13(9):2513-2526.

Pham TV, Henneman AA, Jimenez CR. iq: an R package to estimate relative protein abundances from ion quantification in DIA-MS-based proteomics. Bioinformatics 2020 Apr 15;36(8):2611-2613.

The meanInt algorithm for protein quantification

Description

Estimates protein abundances by averaging all associated ion intensities

Usage

meanInt(X, aggregation_in_log_space = TRUE)

Arguments

Value

A list of two components is returned

Author(s)

Thang V. Pham

References

Pham TV, Tran CTM, Henneman AA, Pham LHC, Le DG, Can AH, Bui PHL, Piersma SR, Jimenez CR. Boosting the speed and accuracy of protein quantification algorithms in mass spectrometry-based proteomics. Journal of Proteome Research 2026 Feb 6; 25(2):1198-1203.

Pham TV, Henneman AA, Jimenez CR. iq: an R package to estimate relative protein abundances from ion quantification in DIA-MS-based proteomics. Bioinformatics 2020 Apr 15;36(8):2611-2613.

A wrapper for the R implementation of the median polish algorithm

Description

Estimates protein abundances using the Tukey median polish algorithm.

Usage

median_polish(X)

Arguments

Value

A list of two components is returned

Author(s)

Thang V. Pham

References

Pham TV, Tran CTM, Henneman AA, Pham LHC, Le DG, Can AH, Bui PHL, Piersma SR, Jimenez CR. Boosting the speed and accuracy of protein quantification algorithms in mass spectrometry-based proteomics. Journal of Proteome Research 2026 Feb 6; 25(2):1198-1203.

Pham TV, Henneman AA, Jimenez CR. iq: an R package to estimate relative protein abundances from ion quantification in DIA-MS-based proteomics. Bioinformatics 2020 Apr 15;36(8):2611-2613.

Tukey JW. Exploratory Data Analysis, Reading Massachusetts: Addison-Wesley, 1977.

Plotting the underlying quantitative data for a protein

Description

Displays the underlying data for a protein.

Usage

plot_protein(X, main = "", col = NULL, split = 0.6, ...)

Arguments

Value

A NULL value is returned.

Author(s)

Thang V. Pham

References

Pham TV, Henneman AA, Jimenez CR. iq: an R package to estimate relative protein abundances from ion quantification in DIA-MS-based proteomics. Bioinformatics 2020 Apr 15;36(8):2611-2613.

Examples

data("spikeins")
head(spikeins)
# This example set of spike-in proteins has been 'median-normalized'.
norm_data <- iq::preprocess(spikeins, median_normalization = FALSE, pdf_out = NULL)
protein_list <- iq::create_protein_list(norm_data)
iq::plot_protein(protein_list$P00366, main = "Protein P00366", split = NULL)

Data preprocessing for protein quantification

Description

Prepares a long-format input including removing low-intensity ions and performing median normalization.

Usage

preprocess(quant_table,
           primary_id = "PG.ProteinGroups",
           secondary_id = c("EG.ModifiedSequence", "FG.Charge", "F.FrgIon", "F.Charge"),
           sample_id = "R.Condition",
           intensity_col = "F.PeakArea",
           median_normalization = TRUE,
           log2_intensity_cutoff = 0,
           pdf_out = "qc-plots.pdf",
           pdf_width = 12,
           pdf_height = 8,
           intensity_col_sep = NULL,
           intensity_col_id = NULL,
           na_string = "0",
           show_boxplot = TRUE,
           verbose = TRUE)

Arguments

Details

When entries in the intensity column contain multiple values, this function will replicate entries in other column and the secondary_id will be appended with corresponding entries in intensity_col_id when it is provided. Otherwise, integer values 1, 2, 3, etc... will be used.

Value

A data frame is returned with following components

Author(s)

Thang V. Pham

References

Pham TV, Tran CTM, Henneman AA, Pham LHC, Le DG, Can AH, Bui PHL, Piersma SR, Jimenez CR. Boosting the speed and accuracy of protein quantification algorithms in mass spectrometry-based proteomics. Journal of Proteome Research 2026 Feb 6; 25(2):1198-1203.

Pham TV, Henneman AA, Jimenez CR. iq: an R package to estimate relative protein abundances from ion quantification in DIA-MS-based proteomics. Bioinformatics 2020 Apr 15;36(8):2611-2613.

Examples

data("spikeins")
head(spikeins)
# This example set of spike-in proteins has been 'median-normalized'.
norm_data <- iq::preprocess(spikeins, median_normalization = FALSE, pdf_out = NULL)

Protein quantification from a long-format table

Description

Reads a long-format table of ion intensities (a tab-separated text file, a parquet file or a data frame), filters and normalizes the data, quantifies all proteins with one of the methods of process_matrix, and writes a table with one row per protein and one column per sample.

Usage

process(input_data,
        output_filename,
        sample_id = "Run",
        primary_id = "Protein.Group",
        secondary_id = "Precursor.Id",
        intensity_col = "Precursor.Normalised",
        annotation_col = NULL,
        filter_string_equal = NULL,
        filter_string_not_equal = NULL,
        filter_double_less = c("Q.Value" = "0.01", "PG.Q.Value" = "0.01"),
        filter_double_greater = NULL,
        intensity_col_sep = NULL,
        intensity_col_id = NULL,
        na_string = "0",
        normalization = "median",
        log2_intensity_cutoff = 0,
        pdf_out = "qc-plots.pdf", pdf_width = 12, pdf_height = 8,
        show_boxplot = TRUE, boxplot_max_points = 10000,
        peptide_extractor = NULL,
        rfasta = NULL,
        method = "maxlfq_bit",
        p1 = NULL, p2 = NULL, k = 1.345, min_M = 15,
        rescale_method = "to_mean",
        n_threads = -1,
        verbose = TRUE)

Arguments

Details

Input. The defaults are the same for all inputs: the precursor quantities Precursor.Normalised per Run of a DIA-NN report. A data frame is read in place, without copying (fast_read_data_frame). A parquet file is read one row group at a time, only the needed columns (fast_read_parquet; requires the arrow package). Any other input is read as a tab-separated text file (fast_read). Filtering is the same for all inputs; a missing filter value (NA, NaN, empty, or text that is not a number) keeps the row.

Preprocessing. Intensities are log2-transformed; zero and negative intensities are removed, as are values not above log2_intensity_cutoff (by default 0, i.e. intensities of 1 or less); and samples are median normalized (fast_preprocess).

Quantification. For each protein, a matrix of log2 intensities is formed with ions as rows (in input order) and all samples as columns; an ion with more than one value in a sample gets the log2 of the summed intensities. The matrix is quantified with method and the result rescaled with rescale_method, as in process_iq_format, without writing iq-format files. For the built-in methods, all proteins are processed in C++ in parallel. A function given as method is called in R for each protein with this matrix, with the ion identifiers as row names and the sample names as column names, and must return a numeric vector with one value per sample. Values that are not finite (-Inf, Inf, NaN) become NA. For example, function(X) log2(colSums(2^X, na.rm = TRUE)) with rescale_method = NULL gives the log2 summed intensities per sample, with NA for a sample without data, as the method "sum".

Output. A tab-separated file with one row for each protein that has data after preprocessing. The columns are primary_id, the annotation_col columns, the fragment statistics (if peptide_extractor is given), and one column per sample with the log2 estimates; a sample without data for a protein is NA.

Fragment statistics. If peptide_extractor is not NULL, the columns n_fragments (the number of MS2 fragment ions or precursors depending on intensity_col), n_peptides (the number of unique values of peptide_extractor on the ions), peptides and XIC (the summed intensities, before preprocessing) are added, and coverage (sequence coverage in percent for each protein in the group) if rfasta is given. peptide_extractor is called with the secondary_id part of the ion identifiers, without the suffix of the intensity values (1, 2, ... or the intensity_col_id values), e.g. a precursor id such as (UniMod:1)AAAAAAGAGPEMVR2, also when intensity_col has one column per fragment ion; the ion_prefix component returned by fast_read, fast_read_parquet or fast_read_data_frame shows these values. An example extractor for DIA-NN precursor ids is function(x) gsub("\\(UniMod:\\d+\\)|\\d", "", x), which removes the modifications and the charge state: (UniMod:1)AAAAAAGAGPEMVR2 becomes AAAAAAGAGPEMVR.

Value

NULL, invisibly. The result is written to output_filename.

Author(s)

Thang V. Pham

References

Pham TV, Tran CTM, Henneman AA, Pham LHC, Le DG, Can AH, Bui PHL, Piersma SR, Jimenez CR. Boosting the speed and accuracy of protein quantification algorithms in mass spectrometry-based proteomics. Journal of Proteome Research 2026 Feb 6; 25(2):1198-1203.

See Also

fast_read, fast_read_parquet, fast_read_data_frame, fast_preprocess, process_matrix, rescale, process_iq_format, read_fasta_file

Process data in the iq format

Description

Estimates protein abundance of all proteins in the iq format data.

Usage

process_iq_format(iqformat_folder,
                  output_filename = NULL,
                  method = "maxlfq_bit", 
                  p1 = NULL, p2 = NULL, k = 1.345, min_M = 15,
                  rescale_method = "to_mean",
                  buffer_mb = 1024,
                  n_threads = -1,
                  verbose = TRUE)

Arguments

Details

The protein files are read in batches of about buffer_mb megabytes of values, and the proteins of a batch are quantified in parallel (n_threads), as in process. With buffer_mb = NULL, or a function as method, each protein is read with read.delim and quantified with process_matrix in R. The log2 values are shifted by the normalization of each sample (the normalization column of _sample.tsv, if present. All columns of _protein.tsv are written before the samples: the protein annotation, and the fragment statistics if long_format_to_iq_format computed them.

Value

NULL, invisibly. A tab-separated table is written to output_filename (by default iqformat_folder followed by -quant.tsv), with one row for each protein with values, columns for the protein annotation and statistics, and the samples.

Author(s)

Thang V. Pham

References

Pham TV, Tran CTM, Henneman AA, Pham LHC, Le DG, Can AH, Bui PHL, Piersma SR, Jimenez CR. Boosting the speed and accuracy of protein quantification algorithms in mass spectrometry-based proteomics. Journal of Proteome Research 2026 Feb 6; 25(2):1198-1203.

See Also

long_format_to_iq_format, process, process_matrix, rescale

Long format to a wide format table using the MaxLFQ algorithm

Description

A convenient function combining multiple steps to process a long format table using the MaxLFQ algorithm.

Usage

process_long_format(input_data,
                    output_filename,
                    sample_id = "File.Name",
                    primary_id = "Protein.Group",
                    secondary_id = "Precursor.Id",
                    intensity_col = "Fragment.Quant.Corrected",
                    annotation_col = NULL,
                    filter_string_equal = NULL,
                    filter_string_not_equal = NULL,
                    filter_double_less = c("Q.Value" = "0.01", "PG.Q.Value" = "0.01"),
                    filter_double_greater = NULL,
                    intensity_col_sep = ";",
                    intensity_col_id = NULL,
                    na_string = "0",
                    normalization = "median",
                    log2_intensity_cutoff = 0,
                    pdf_out = "qc-plots.pdf",
                    pdf_width = 12,
                    pdf_height = 8,
                    show_boxplot = TRUE,
                    peptide_extractor = NULL,
                    rfasta = NULL,
                    verbose = TRUE)

Arguments

Value

An input data frame is filtered and processed with preprocess and fast_MaxLFQ; an input file is processed with fast_read, fast_preprocess, and fast_MaxLFQ. Subsequently, the result is written to output_filename. The quantification values are in log2 space. A NULL value is returned. If peptide_extractor is not NULL, fragment statistics for each protein will be calculated based on the result of the extractor function: the columns n_fragments and n_peptides, and for an input file also peptides, XIC (the summed intensities, before preprocessing) and, with rfasta, coverage. Counting the number of peptides contributing to a protein is possible using an appropriate extractor function. An example value for peptide_extractor is function(x) gsub("\\(UniMod:\\d+\\)|\\d|_", "", x), which removes modifications, digit numbers, and underscores in an ion descriptor to obtain peptide sequences. One can examine the ion component returned by the fast_read function to derive a regular expression to be used in the gsub function above.

Author(s)

Thang V. Pham

References

Pham TV, Tran CTM, Henneman AA, Pham LHC, Le DG, Can AH, Bui PHL, Piersma SR, Jimenez CR. Boosting the speed and accuracy of protein quantification algorithms in mass spectrometry-based proteomics. Journal of Proteome Research 2026 Feb 6; 25(2):1198-1203.

Pham TV, Henneman AA, Jimenez CR. iq: an R package to estimate relative protein abundances from ion quantification in DIA-MS-based proteomics. Bioinformatics 2020 Apr 15;36(8):2611-2613.

See Also

fast_read, fast_preprocess, fast_MaxLFQ

Process individual matrix

Description

Estimates a row vector best representing all rows in the input numeric matrix.

Usage

process_matrix(X, 
               method = "maxlfq_bit", 
               p1 = NULL, p2 = NULL, k = 1.345, min_M = 15,
               rescale_method = "to_mean",
               n_threads = -1)

Arguments

Details

The methods "sum", "mean" and "topN" summarize the ions of each sample directly: the log2 of the summed intensities, the mean of the log2 intensities, and the mean of the log2 intensities of the p2 ions with the highest mean log2 intensity over the samples (as topN). A sample without data gets NA. To keep these values as they are, use rescale_method = NULL; the default rescale_method shifts them like the estimates of the other methods.

The methods "maxlfq" and "weighted_maxlfq" solve the MaxLFQ least-squares problem with the iterative solver of "maxlfq_bit". A package built with the original MaxLFQ implementation (using the Eigen library, see the file src/Makevars of the source package) solves it exactly instead.

Value

A numeric row vector, one value per sample (column of X). For the built-in methods, a matrix with a single row gives that row (before rescaling; the "to_mean" and "to_sum" rescaling leave it unchanged).

Author(s)

Thang V. Pham

References

Pham TV, Tran CTM, Henneman AA, Pham LHC, Le DG, Can AH, Bui PHL, Piersma SR, Jimenez CR. Boosting the speed and accuracy of protein quantification algorithms in mass spectrometry-based proteomics. Journal of Proteome Research 2026 Feb 6; 25(2):1198-1203.

Examples

# a data matrix for a single protein: log2 intensities, ions in rows, samples in columns
X <- matrix(c(10, 12, 12, NA, NA, 10), nrow = 2,
            dimnames = list(c("ion1", "ion2"),
                            c("sample1", "sample2", "sample3")))
X

# relative quantification: sample2 is 4-fold higher than sample1 (a difference of 2 in
# log2 space), the difference of ion1, the only ion they share; similarly, sample1 is
# 4-fold higher than sample3 (ion2)
process_matrix(X, method = "maxlfq_bit", rescale_method = NULL)  # 11 13 9

# rescaled to the total intensity: the same differences between samples, and the
# intensities of the estimates add up to those of the data
x <- process_matrix(X, method = "maxlfq_bit", rescale_method = "to_sum")
x                       # 10.93 12.93 8.93
sum(2^X, na.rm = TRUE)  # 10240
sum(2^x)                # 10240

Merging rows with identical values in a particular column in a table

Description

Collapses rows with identical values in a particular column in a table. When the values in each row are proportional such as intensities of multiple fragments of a protein, the MaxLFQ algorithm is recommended.

Usage

process_wide_format(input_filename,
                    output_filename,
                    id_column,
                    quant_columns,
                    data_in_log_space = FALSE,
                    annotation_columns = NULL,
                    method = "maxLFQ")

Arguments

Details

Method "maxLFQ_R" implements the MaxLFQ algorithm pure R. It is slower than "maxLFQ".

Method "maxInt" selects row with maximum intensity (top 1).

Method "sum" sum all intensities.

Method "least_na" selects row with the least number of missing values.

The value of method can be a function such as function(x) log2(colSums(2^x, na.rm = TRUE)) for summing all intensities in the original space.

Value

The result table is written to output_filename. A NULL value is returned.

Author(s)

Thang V. Pham

References

Pham TV, Tran CTM, Henneman AA, Pham LHC, Le DG, Can AH, Bui PHL, Piersma SR, Jimenez CR. Boosting the speed and accuracy of protein quantification algorithms in mass spectrometry-based proteomics. Journal of Proteome Research 2026 Feb 6; 25(2):1198-1203.

Pham TV, Henneman AA, Jimenez CR. iq: an R package to estimate relative protein abundances from ion quantification in DIA-MS-based proteomics. Bioinformatics 2020 Apr 15;36(8):2611-2613.

Reading a FASTA file

Description

Reads protein sequences from a FASTA file into a named list, for example for the rfasta argument of process and process_long_format.

Usage

read_fasta_file(fasta_file_name, regex_id = "^>.*\\|(.*)\\|.*")

Arguments

Details

A sequence is the concatenation of the lines below its header, with all white space removed. A header that does not match regex_id gives the whole header line, without >, as the name. Lines before the first header are ignored.

Value

A named list of sequences (character strings), in the order of the file.

Author(s)

Thang V. Pham

References

Pham TV, Tran CTM, Henneman AA, Pham LHC, Le DG, Can AH, Bui PHL, Piersma SR, Jimenez CR. Boosting the speed and accuracy of protein quantification algorithms in mass spectrometry-based proteomics. Journal of Proteome Research 2026 Feb 6; 25(2):1198-1203.

See Also

process

Rescale relative quantification

Description

Each connected component of the input values can be rescaled to match the level of the overall input intensities.

Usage

rescale(x, X, method)

Arguments

Value

A row vector of rescaled values of the same size as x, with each connected component rescaled independently. For the "median_mean" method, single-sample components are rescaled to the median of input data, while all other components are rescaled to the mean. For the "to_median" and "to_mean" methods, each component is rescaled to the median or mean of the input data. For the "to_sum" method, each component is shifted so that its values, in linear scale (2^x), add up to the summed intensities of its input data (2^X) in the samples of the component; a single-sample component gets the log2 of the summed intensities of that sample. For all methods, a sample without input data (an all-NA column of X) gets NA, and a component whose values in x are all NA is left unchanged.

Author(s)

Thang V. Pham

References

Pham TV, Tran CTM, Henneman AA, Pham LHC, Le DG, Can AH, Bui PHL, Piersma SR, Jimenez CR. Boosting the speed and accuracy of protein quantification algorithms in mass spectrometry-based proteomics. Journal of Proteome Research 2026 Feb 6; 25(2):1198-1203.

An example dataset of 12 spike-in proteins

Description

A subset of the Bruderer 2015 dataset containing 12 spike-in proteins. The full dataset was exported from the Spectronaut software. The complete dataset has been median-normalized.

Usage

data("spikeins")

Format

A data frame with 18189 observations on the following 9 variables.

R.Condition

Sample names.

PG.ProteinGroups

Protein identifiers.

EG.ModifiedSequence

Sequence of the fragment ions.

FG.Charge

Fragment group charge.

F.FrgIon

Fragment ions.

F.Charge

Fragment charges.

F.PeakArea

Quantitative values.

PG.Genes

Gene names.

PG.ProteinNames

Protein names.

Examples

data("spikeins")
head(spikeins)

The topN algorithm for protein quantification

Description

Estimates protein abundances using the N most intense ions.

Usage

topN(X, N = 3, aggregation_in_log_space = TRUE)

Arguments

Value

A list of two components is returned

Author(s)

Thang V. Pham

References

Pham TV, Henneman AA, Jimenez CR. iq: an R package to estimate relative protein abundances from ion quantification in DIA-MS-based proteomics. Bioinformatics 2020 Apr 15;36(8):2611-2613.