1 OVERVIEW

artMS is a Bioconductor package that provides a set of tools for the analysis and integration of large-scale proteomics (mass-spectrometry-based) datasets obtained using the popular proteomics software MaxQuant.

artMS also perfoms basic quality control and relative quantification for metabolomics datasets obtained using the alignment table generated by MarkerView.

The functions available in artMS can be grouped into the following categories:

  • Multiple quality control (QC) functions.
  • Relative quantification using MSstats.
  • Downstream analysis and integration of quantifications (enrichment, clustering, PCA, summary plots, etc)
  • Generation of input files for other tools, including SAINTq, SAINTexpress, Photon, and Phosfate

1.1 How to install

Before you begin, ensure that your system is running an R version >= 3.6 or the installation of artMS won’t work. You can check the R version running on your system by executing the function getRversion()

If the outcome is >= 3.6.0, congratulations! you can move forward. If it is not, then you need to install the latest version of R in your system.

Two options to install artMS:

  • Official BioConductor releases (recommended)

artMS is available in BioConductor. Instructions to install the package are available here.

(Why Bioconductor? Here you can find a nice summary of many good reasons).

  • Development version from Github

(Warning: not stable, but it has the latest)

Assuming that you have an R (>= 3.6) version running on your system, follow these steps:

install.packages("devtools")
library(devtools)
install_github("biodavidjm/artMS")

Once installed, the package can be loaded and attached to your current workspace as follows:

library(artMS)

1.2 Input files

artMS performs the different analyses taking as input the following files:

  • evidence.txt file: output of the quantitative proteomics software package MaxQuant.
  • keys.txt (tab-delimited) txt file generated by the user describing the experimental design.
  • contrast.txt (tab-delimited) txt file generated by the user with the comparisons between conditions to be quantified.

Check below to find out more about generating the input files.

1.3 Configuration file

artmsQuantification() requires a large number of arguments, specially those related to the statistical package MSstats. To facilite the task of providing all those arguments, the function artmsQuantification() takes a config file (in yaml format) for the customization of the parameters for quantification (using MSstats) and other operations, including QC analyses, charts, and annotations.

A configuration file template can be generated by running artmsWriteConfigYamlFile()

Check below to learn the details of the configuration file.

1.4 Basic workflows

1.4.1 Proteomics

  • Generate the input files: Check the input files section for details

  • Quality Control: if you are interested in performing only quality control analysis, run the following functions:

    • artmsQualityControlEvidenceBasic(): QC based on the evidence.txt file
    • artmsQualityControlEvidenceExtended(): based on the evidence.txt file
    • artmsQualityControlSummaryExtended(): based on the summary.txt file

  • Relative Quantification: fill up the configuration file and run the following function:

  • Analysis of Quantifications: performs annotations, clustering analysis, PCA analysis, enrichment analysis by running the function

  • Miscellaneous functions: Check below to discover more useful functions provided by the artMS package.

1.4.2 Metabolomics

artMS also enables the relative quantification of untargeted polar metabolites using the alignment table generated by MarkerView. This means that the metabolites do not need to have an ID, as the m/z and retention time will be used as identifiers. Typical workflow:

  • Run QC on the metabolomics dataset: artmsQualityControlMetabolomics()

  • Relative quantification: artmsQuantification() (notice that a few options must be changed in the config file before running the function)

Please, keep in mind that most of the functions won’t work for metabolomics data due to annotation issues (protein/gene ids are the primary ids for most of the functions). Check the metabolomics section to find out more.

2 REQUIRED INPUT FILES

2.1 Input files

Three basic (tab-delimited) files are required to perform the full pack of operations:

2.1.1 evidence.txt

The output of the quantitative proteomics software package MaxQuant. It combines all the information about the identified peptides.

2.1.2 keys.txt

Tab delimited file generated by the user. It summarizes the experimental design of the evidence file. artMS merges the keys.txt and evidence.txt by the “RawFile” column. Each RawFile corresponds to a unique individual experimental technical replicate / biological replicate / Condition / Run.

For any basic label-free proteomics experiment, the keys file must contain the following columns and rules:

  • RawFile: The name of the RAW-file for which the mass spectral data was derived from.
  • IsotopeLabelType: 'L' for label free experiments ('H' will be used for SILAC experiments, see below)
  • Condition: The conditions names must follow these rules:
    • Use only letters (A - Z, both uppercase and lowercase) and numbers (0 - 9). The only special character allowed is underscore (_).
    • Very important: A condition name cannot begin with a number (R limitation).
  • BioReplicate: biological replicate number. It is based on the condition name. Use as prefix the corresponding Condition name, and add as suffix dash (-) plus the biological replicate number. For example, if condition H1N1_06H has too biological replicates, name them H1N1_06H-1 and H1N1_06H-2
  • Run: a unique number for all the MS runs (from 1 to the total number of raw files). It will be specially useful when having technical replicates. A special case is SILAC experiments (H and L label are run simultaneously. See below to find out more)

Example of keys file: check the artMS data object artms_data_ph_keys

RawFile IsotopeLabelType Condition BioReplicate Run
qx006145 L Cal_33 Cal_33-1 1
qx006148 L Cal_33 Cal_33-4 4
qx006151 L HSC6 HSC6-2 6
qx006152 L HSC6 HSC6-3 7

Tip: it is recommended to use Microsoft Excel (OpenOffice Cal / or similar) to generate the keys file. Do not forget to choose the format = Tab Delimited Text (.txt) when saving the file (use save as option)

2.1.3 contrast.txt

The comparisons between conditions that the user wants to quantify.

  • Example #1: the comparison for the keys described above would be:
HSC6-Cal_33
  • Example #2: let’s quantify changes in protein abundance between wild type (WT_A549) relative to two additional experimental conditions with drugs (WT_DRUG_A and WT_DRUG_B), but also changes in protein abundance between DRUG_A and DRUG_B, the contrast file would look like this:
WT_DRUG_A-WT_A549
WT_DRUG_B-WT_A549
WT_DRUG_A-WT_DRUG_B

Requirements:

  • The two conditions to be compared must be separated by a dash symbol (-), and only one dash symbol is allowed, i.e., only one comparison per line.

As a result of the quantification, the condition on the left will take the positive log2FC sign -if the protein is more abundant in condition on the left (numerator), and the condition on the right the negative log2FC -if a protein is more abundant in condition on the right term (denominator).

Example of wrong comparisons

Only condition names are allowed. Individual Bioreplicates cannot be compared. For example, this is wrong:

HSC6-Cal_33-1

2.2 The artMS configuration file

The configuration file (in yaml format) contains a variety of options available for the QC, quantification, and annotations performed by artMS.

To generate a sample configuration file, go to the project folder (setwd(/path/to/your/working/folder/)) and execute:

library(artMS)
## Warning: replacing previous import 'MASS::select' by 'dplyr::select' when
## loading 'MSstats'
## 
## 
## Registered S3 method overwritten by 'xts':
##   method     from
##   as.zoo.xts zoo
artmsWriteConfigYamlFile(config_file_name = "config.yaml", 
                         verbose = FALSE)

Open the config.yaml file with your favorite editor (RStudio for example). Although it might look complex, the default options work very well.

The configuration (yaml) file contains the following sections:

2.2.1 Section: files

files :
  evidence : /path/to/the/evidence.txt
  keys : /path/to/the/keys.txt
  contrasts : /path/to/the/contrast.txt
  summary: /path/to/the/summary.txt # Optional
  output : /path/to/the/results_folder/ph-results.txt

The file path/name of the required files. It is recommended to create a new folder in your folder project (for example, results_folder). The results file name (e.g. -results.txt) will be used as prefix for the several files (txt and pdf) that will be generated.

2.2.2 Section: qc

qc:
  basic: 1 # 1 = yes; 0 = no
  extended: 1 # 1 = yes; 0 = no
  extendedSummary: 0 # 1 = yes; 0 = no

Select to perform both ‘basic’ and ‘extended’ quality control based on the evidence.txt file or ‘extendedSummary’ based on the summary.txt file. Read below to find out more about the details of each type of analysis.

2.2.3 Section: data

data:
  enabled : 1 # 1 = yes; 0 = no
  fractions: 
    enabled : 0 # 1 for protein fractionation
  silac: 
    enabled : 0 # 1 for SILAC experiments
  filters: 
    enabled : 1
    contaminants : 1
    protein_groups : remove # remove, keep
    modifications : AB # PH, UB, AB, APMS
  sample_plots : 1 # correlation plots

Let’s break it down data:

  • enabled : 1: to pre-process the data provided in the files section. 0: won’t process the data (and a pre-generated MSstats file will be expected)

  • fractions: Multiple fractionation or separation methods are often combined in proteomics to improve signal-to-noise and proteome coverage and to reduce interference between peptides in quantitative proteomics.
    • enabled : 1 for fractionation dataset. See Special case: Protein Fractionation below for details
    • enabled : 0 no fractions
  • silac:
    • enabled : 1: check if the files belong to a SILAC experiment. See Special case: SILAC below for details
    • enabled : 0: no silac experiment (default)
  • filters:
    • enabled : 1 Enables filtering (this section)
    • contaminants : 1 Removes contaminants (CON__ and REV__ labeled by MaxQuant)
    • protein_groups : remove choose whether remove or keep protein groups
    • modifications : AB any of the proteomics experiments, PH, UB, or AC for posttranslational modifications, AB or APMS otherwise.
  • sample_plots
    • 1 Generate correlation plots
    • 0 otherwise

2.2.4 Section: MSstats

msstats :
  enabled : 1
  msstats_input : # `-mss.txt` file or blank (default)
  profilePlots : none 
  normalization_method : equalizeMedians
  normalization_reference :  # blank (default) if equalizeMedians
  summaryMethod : TMP 
  censoredInt : NA  
  cutoffCensored : minFeature  
  MBimpute : 1 
  feature_subset: all

Let’s break it down:

  • enabled : Choose 1 to run MSstats, 0 otherwise.
  • msstats_input : leave it blank if MSstats will be run (previous enabled : 1). But if MSstats was already run and the evidence-mss.txt file is available, then choose enabled : 0 and provide here the evidence-mss.txt file path/name
  • profilePlots : Choose one of the following options:
    • before plot before normalization
    • after plot after normalization
    • before-after: recommended, although computational expensive
    • none no normalization plots
  • normalization_method : available options:
    • equalizeMedians
    • quantile
    • 0: no normalization (not recommended)
    • globalStandards if selected, specified the reference protein in normalization_reference (next)
  • normalization_reference : UniProt id if globalStandards is chosen as the normalization_method (above)
  • summaryMethod : TMP # “TMP”(default) means Tukey’s median polish, which is robust estimation method. “linear” uses linear mixed model. “logOfSum” conducts log2 (sum of intensities) per run.
  • censoredInt :
    • NA (default) Missing values are censored or at random. ‘NA’ assumes that all ‘NA’s in ’Intensity’ column are censored.
    • 0 uses zero intensities as censored intensity. In this case, NA intensities are missing at random. The output from Skyline should use 0. Null assumes that all NA intensities are randomly missing.
  • cutoffCensored :
    • minFeature Cutoff value for censoring. Only with censoredInt : NA or 0. Default is ‘minFeature’, which uses minimum value for each feature.
    • minFeatureNRun uses the smallest between minimum value of corresponding feature and minimum value of corresponding run.
    • minRun uses minimum value for each run.
  • MBimpute :
    • TRUE only for summaryMethod="TMP" and censoredInt='NA' or 0. TRUE (default) imputes ‘NA’ or ‘0’ (depending on censoredInt option) by Accelerated failure model.
    • FALSE uses the values assigned by cutoffCensored.
  • feature_subset :
    • all : default
    • highQuality : this option seems to be buggy right now

Check MSstats documentation to find out more about every option.

2.2.5 Section: output_extras

  enabled : 1 # if 0, won't process anything on this section
  annotate :  
    enabled: 1 
    species : HUMAN
  plots:
    volcano: 1
    heatmap: 1
    LFC : -1.5 1.5 # Range of minimal log2fc
    FDR : 0.05
    heatmap_cluster_cols : 0
    heatmap_display : log2FC # log2FC or pvalue

Extra actions to perform based on the MSstats results, including annotations and plots (heatmaps and volcano plots). Let’s break it down:

  • enabled : 1 (default) enables this section, 0 turns it off
  • annotate :
    • enabled: 1 (default), will generate a -results-annotated.txt file that includes Gene and Protein.Name (only for supported species)
    • species: The supported species are: HUMAN, MOUSE, ANOPHELES, ARABIDOPSIS, BOVINE, WORM, CANINE, FLY, ZEBRAFISH, ECOLI_STRAIN_K12, ECOLI_STRAIN_SAKAI, CHICKEN, RHESUS, MALARIA, CHIMP, RAT, YEAST, PIG, XENOPUS
  • plots : options for additional plots
    • volcano : 1
    • LFC : log2 fold change cutoff (minimal negative and positive value)
    • FDR : false discovery rate cutoff for significance (recommended: 0.05)
    • heatmap : correlation plots
    • heatmap_cluster_cols : 1 perfoms clustering of columns, 0 (default) doesn’t
    • heatmap_display : choose to display either log2FC or pvalue

2.3 Special case: Protein fractionation

To handle protein fractionation experiments, two options must be activated

  1. keys.txt: The keys file must contain an additional column named “FractionKey” with the information about fractions. For example:
Raw.file IsotopeLabelType Condition BioReplicate Run FractionKey
S9524_Fx1 L AB AB-1 1 1
S9524_Fx2 L AB AB-1 1 2
S9524_Fx3 L AB AB-1 1 3
S9524_Fx4 L AB AB-1 1 4
S9524_Fx5 L AB AB-1 1 5
S9524_Fx6 L AB AB-1 1 6
S9524_Fx7 L AB AB-1 1 7
S9524_Fx8 L AB AB-1 1 8
S9524_Fx9 L AB AB-1 1 9
S9524_Fx10 L AB AB-1 1 10
S9525_Fx1 L AB AB-2 2 1
S9525_Fx2 L AB AB-2 2 2
S9525_Fx3 L AB AB-2 2 3
S9525_Fx4 L AB AB-2 2 4
S9525_Fx5 L AB AB-2 2 5
S9525_Fx6 L AB AB-2 2 6
S9525_Fx7 L AB AB-2 2 7
S9525_Fx8 L AB AB-2 2 8
S9525_Fx9 L AB AB-2 2 9
S9525_Fx10 L AB AB-2 2 10
S9526_Fx1 L AB AB-3 3 1
S9526_Fx2 L AB AB-3 3 2
S9526_Fx3 L AB AB-3 3 3
S9526_Fx4 L AB AB-3 3 4
S9526_Fx5 L AB AB-3 3 5
S9526_Fx6 L AB AB-3 3 6
S9526_Fx7 L AB AB-3 3 7
S9526_Fx8 L AB AB-3 3 8
S9526_Fx9 L AB AB-3 3 9
S9526_Fx10 L AB AB-3 3 10
  1. config.yaml: Enable fractions in the configuration file as follow:
fractions: 
  enabled : 1 # 1 for protein fractions, 0 otherwise

2.4 Special case: SILAC

One of the most widely used techniques that enable relative protein quantification is stable isotope labeling by amino acids in cell culture (SILAC). The keys.txt file can capture the typical SILAC experiment. The following example shows a SILAC experiment with two conditions, two biological replicates, and two technical replicates:

RawFile IsotopeLabelType Condition BioReplicate Run
QE20140321-01 H iso iso-1 1
QE20140321-02 H iso iso-1 2
QE20140321-04 L iso iso-2 3
QE20140321-05 L iso iso-2 4
QE20140321-01 L iso_M iso_M-1 1
QE20140321-02 L iso_M iso_M-1 2
QE20140321-04 H iso_M iso_M-2 3
QE20140321-05 H iso_M iso_M-2 4

It is also required to activate the silac option in the yaml file as follows:

silac: 
  enabled : 1 # 1 for SILAC experiments

3 QUALITY CONTROL

artMS provides 3 functions to perform QC analyses.

3.1 Basic QC (evidence.txt-based)

The basic quality control analysis takes as input both the evidence.txt and keys.txt files and generates several QC plots exploring different aspects of the MS data. Run it as follows:

artmsQualityControlEvidenceBasic(evidence_file = artms_data_ph_evidence,
                                 keys_file = artms_data_ph_keys,
                                 prot_exp = "PH")

The following pdf files are generated by default:

  • -basicReproducibility.pdf: correlation dot plot for all the combinations of biological replicates of every condition, based on MS Intensity values of features (peptide+charge)
  • -correlationMatrixBR.pdf: It contains 3 pages. Correlation matrix for all the biological replicates using MS Intensity values, Clustering matrix of the MS Intensities, and correlation distribution histogram.
  • -correlationMatrixBR.pdf: Same as the previous one, but based on MS Intensity values of Conditions
  • -IntensityDistributions.pdf: 2 pages. Box-dot plot and Jitter plot of MS (raw) intensity values for each biological replicate.
  • -intensityStats.pdf: several pages, including bar plots of Total Sum of Intensities in BioReplicates, Total Sum of Intensities in Conditions, Total Peptide Counts in BioReplicates, Total Peptide Counts in conditions grouped by categories (CON: contaminants, PROT peptides, REV reversed sequences used by MaxQuant to estimate the FDR); Box plots of MS Intensity values per biological replicates and conditions; bar plots of total intensity (excluding contaminants) by bioreplicates and conditions; Bar plots of total feature counts by bioreplicates and conditions.
  • -ptmStats.pdf: If any PTM is selected (PH, UB, AC) an extra pdf file will be generated with stats related to the selected modification, including: bar plot of peptide counts and intensities, broken by PTM/other categories; bar plots of total sum-up of MS intensity values by other/PTM categories.

Check ?artmsQualityControlEvidenceBasic() to find out more options about this function.

Next, for illustration purposes, let’s show how to generate only one plot (e.g. INTDIST):

# But for illustration purposes printing only INTDIST plot:
library(artMS)
suppressWarnings(
artmsQualityControlEvidenceBasic(evidence_file = artms_data_ph_evidence,
                                 keys_file = artms_data_ph_keys,
                                 prot_exp = "PH",
                                 plotINTDIST = TRUE,
                                 plotREPRO = FALSE,
                                 plotCORMAT = FALSE,
                                 plotINTMISC = FALSE,
                                 plotPTMSTATS = FALSE,
                                 printPDF = FALSE,
                                 verbose = FALSE))

3.2 Extended QC (evidence.txt-based)

It takes as input the evidence.txt and keys.txt files as follows:

artmsQualityControlEvidenceExtended(evidence_file = artms_data_ph_evidence,
                                    keys_file = artms_data_ph_keys)

and generates the following QC plots:

  • plotPSM (QC_Plots_PSM.pdf) : it generates peptide-spectrum-matches (PSMs) statistics plot: Page 1 shows the number of PSMs confidently identified in each BioReplicate. If replicates are present, Page 2 shows the mean number of PSMs per condition with error bar showing the standard error of the mean. Note that potential contaminant proteins are plotted separately.
  • plotIONS (QC_Plots_IONS.pdf) : it generates peptide ion statistics plot: A peptide ion is defined in the context of m/z, in other words, an unique peptide sequence may give rise to multiple ions with different charge state and/or amino acid modification. Page 1 shows the number of ions confidently identified in each BioReplicate . If replicates are present, Page 2 shows the mean number of peptide ions per condition with error bar showing the standard error of the mean. Note that potential contaminant proteins are plotted separately.
  • plotTYPE (QC_Plots_TYPE.pdf): it generates identification type statistics plot: MaxQuant classifies each peptide identification into different categories (e.g., MSMS, MULTI-MSMS, MULTI-SECPEP). Page 1 shows the distribution of identification type in each BioReplicate
  • plotPEPTIDES (QC_Plots_PEPTIDES.pdf): it generates peptide statistics plot: Page 1 shows the number of unique peptide sequences (disregard the charge state or amino acid modifications) confidently identified in each BioReplicate. If replicates are present, Page 2 shows the mean number of peptides per condition with error bar showing the standard error of the mean. Note that potential contaminant proteins are plotted separately. Pages 3 and 4 show peptide identification intersection between BioReplicates (the bars are ordered by degree or frequency, respectively), and Page 4 shows the intersections across conditions instead of BioReplicates.
  • plotPROTEINS (QC_Plots_PROTEINS.pdf): it generates protein statistics plot: Page 1 shows the number of protein groups confidently identified in each BioReplicate. If replicates are present, Page 2 shows the mean number of protein groups per condition with error bar showing the standard error of the mean. Note that potential contaminant proteins are plotted separately. Pages 3 and 4 show peptide identification intersection between BioReplicates (the bars are ordered by degree or frequency, respectively), and Page 4 shows the intersections across conditions instead of BioReplicates.
  • plotPIO (QC_Plots_PepIonOversampling.pdf): it generates oversampling statistics plot: Page 1 shows the proportion of all peptide ions (including peptides matched across runs) fragmented once, twice and thrice or more. Page 2 shows the proportion of peptide ions (with intensity detected) fragmented once, twice and thrice or more. Page 3 shows the proportion of peptide ions (with intensity detected and MS/MS identification) fragmented once, twice and thrice or more
  • plotCS (QC_Plots_CHARGESTATE.pdf): it generates charge state plot: Page 1 shows the charge state distribution of PSMs confidently identified in each BioReplicate.
  • plotME (QC_Plots_MASSERROR.pdf): it generates precursor mass error plot: Page 1 shows the distribution of precursor error for all PSMs confidently identified in each BioReplicate.
  • plotMOCD (QC_Plots_MZ.pdf): it generates precursor mass-over-charge plot: Page 1 shows the distribution of precursor mass-over-charge for all PSMs confidently identified in each BioReplicate.
  • plotPEPICV (QC_Plots_PEPINT.pdf): it generates peptide intensity coefficient of variance (CV) plot: The CV is calculated for each feature (peptide ion) identified in more than one replicate. Page 1 shows the distribution of CV’s for each condition, while Page 2 shows the distribution of CV’s within 4 bins of intensity (i.e., 4 quantiles of average intensity).
  • plotPEPDETECT (QC_Plots_PepDetect.pdf): it generates peptide detection frequency plot: Page 1 summarizes the frequency that each peptide is detected across BioReplicates of each condition, showing the percentage of peptides detected once, twice, thrice, and so on (for whatever number of replicates each condition has).
  • plotPROTICV (QC_Plots_ProtInt.pdf): it generates protein intensity coefficient of variance (CV) plot: The CV is calculated for each protein (after summing the peptide intensities) identified in more than one replicate. Page 1 shows the distribution of CV’s for each condition, while Page 2 shows the distribution of CV’s within 4 bins of intensity (i.e., 4 quantiles of average intensity).
  • plotPROTDETECT (QC_Plots_ProtDetect.pdf): it generates protein detection frequency plot: Page 1 summarizes the frequency that each protein group is detected across BioReplicates of each condition, showing the percentage of proteins detected once, twice, thrice, and so on (for whatever number of replicates each condition has). Page 2 shows the feature (peptide ion) intensity distribution within each BioReplicate (potential contaminant proteins are plot separately). Page 3 shows the density of feature intensity for different feature types (i.e., MULTI-MSMS, MULTI-SECPEP).
  • plotIDoverlap (QC-ID-Overlap.pdf): it generates pairwise identification heatmap overlap: Pages 1 and 2 show pairwise peptide and protein overlap between any 2 BioReplicates, respectively.
  • plotIC (QC-IntCorrelation.pdf): it generates pairwise intensity correlation: Page 1 and 3 show pairwise peptide and protein intensity correlation and scatter plot between any 2 BioReplicates, respectively. Page 2 and 4 show principal component analysis at the intensity level for both peptide and proteins, respectively.
  • plotSP (QC-SamplePrep.pdf): it generates sample quality metrics: Page 1 shows missing cleavage distribution of all peptides confidently identified in each BioReplicate. Page 2 shows the fraction of peptides with at least one methionine oxidized in each BioReplicate.

3.3 Extended QC (summary.txt based)

It requires two files:

  • keys.txt file
  • MaxQuant summary.txt file. As described by MaxQuant’s table.pdf, the summary file contains summary information for all the raw files processed with a single MaxQuant run, including statistics on the peak detection. The QC analysis of this file gathers a quick overview on the quality of every RawFile based on this summary.txt. Run it as follows:
artmsQualityControlSummaryExtended(summary_file = "summary.txt",
                                    keys_file = artms_data_ph_keys)

It generates the following pdf plots:

  • plotMS1SCANS: generates MS1 scan counts plot: Page 1 shows the number of MS1 scans in each BioReplicate. If replicates are present, Page 2 shows the mean number of MS1 scans per condition with error bar showing the standard error of the mean. If isFractions is TRUE, each fraction is a stack on the individual bar graphs.

  • plotMS2: generates MS2 scan counts plot: Page 1 shows the number of MSs scans in each BioReplicate. If replicates are present, Page 2 shows the mean number of MS1 scans per condition with error bar showing the standard error of the mean. If isFractions is TRUE, each fraction is a stack on the individual bar graphs.

  • plotMSMS: generates MS2 identification rate (%) plot: Page 1 shows the fraction of MS2 scans confidently identified in each BioReplicate. If replicates are present, Page 2 shows the mean rate of MS2 scans confidently identified per condition with error bar showing the standard error of the mean. If isFractions TRUE, each fraction is a stack on the individual bar graphs.

  • plotISOTOPE: generates Isotope Pattern counts plot: Page 1 shows the number of Isotope Patterns with charge greater than 1 in each BioReplicate. If replicates are present, Page 2 shows the mean number of Isotope Patterns with charge greater than 1 per condition with error bar showing the standard error of the mean. If isFractions TRUE, each fraction is a stack on the individual bar graphs.

4 RELATIVE QUANTIFICATION

The relative quantification is a fundamental step in the analysis of MS data. artMS facilitates and simplifies the analysis using MSstats, a statistical package for the relative quantification of Mass-Spectrometry based proteomics.

All the options and parameters required to run a relative quantification analysis using MSstats (in addition to other options) are summarized in artMS through a configuration file in .yaml format. Check the input-files section to find out more about each of the options.

Different types of proteomics experiments can be quantified including changes in global protein abundance (AB), affinity purification mass spectrometry (APMS), and different type of posttranslational modifications, including phosphorylation (PH), ubiquitination (UB), and acetylation (AC).

artMS also enables the relative quantification of untargeted polar metabolites using the alignment table generated by MarkerView. This means that artMS does not require an ID for the metabolites, as the m/z and retention time will be combined and used as identifiers.

4.1 Quantification of Changes in Global Protein Abundance

The quantification of changes in protein abundance between different conditions requires to fill up the following sections of the config file:

files:
  evidence : /path/to/the/evidence.txt
  keys : /path/to/the/keys.txt
  contrasts : /path/to/the/contrast.txt
  output : /path/to/the/output/results_ptmGlobal/results.txt
  .
  .
  .
data:
  .
  .
  .
  filters:
    modifications : AB

The remaining options can be left unmodified (and run the default parameters). Then run the following artMS function:

artmsQuantification(
  yaml_config_file = '/path/to/config/file/artms_ab_config.yaml')

4.2 Quantification of Changes in Global Phosphorylation or Ubiquitination

Warning: This quantification is only possible for experiments that have used methods to enrich phosphopeptides or ubiquitinated peptides prior to the mass spectrometry analysis.

The global phosphorylation or ubiquitination quantification analysis calculates changes in phosphorylation or ubiquitination at the protein level. This means that all the modified peptides are used to quantify changes in protein phosphorylation or ubiquitination between different conditions. The site-specific analysis (explained next) would quantify changes at the peptide level, i.e., each modified peptide is quantify independently between the different conditions.

Only two sections need to be filled up in the default configuration file:

files:
  evidence : /path/to/the/evidence.txt
  keys : /path/to/the/keys.txt
  contrasts : /path/to/the/contrast.txt
  output : /path/to/the/output/results_ptmGlobal/results.txt
  .
  .
  .
data:
  .
  .
  .
  filters:
    modifications : PH # Use UB for ubiquination

The remaining options can be left unmodified.

Once the configuration yaml file is ready, run the following command:

artmsQuantification(
  yaml_config_file = '/path/to/config/file/artms_phglobal_config.yaml')

4.3 PTM-Site/Peptide-specific Quantification of Changes (PH or UB)

Warning: This quantification is only possible for experiments that have used methods to enrich phosphopeptides or ubiquitinated peptides prior to the mass spectrometry analysis.

The site-specific analysis quantifies changes at the modified peptide level. This means that changes in every modified (PH or UB) peptide of a given protein will be quantified individually. The caveat is that the proportion of missing values should increase relative to the global analysis. Both sites and global ptm analysis are highly correlated due to the fact that only one or two peptides drive the overall changes in PTMs for every protein.

To run a site/peptide specific analysis follow these steps:

  1. Important pre-processing step on the evidence file to enable the ptm-site/peptide-specific analysis. This step takes any of the Proteins id columns selected by the user (either Leading razor protein, Leading protein, or Proteins) and re-annotates it to incorporate the ptm-site/peptide-specific information. By default, this function converts the column Leading razor protein. This step is computational expensive, which means that it might take several minutes to finish (depending on the size of the fasta database, evidence file, computer power, etc)

It also requires the same reference proteome (fasta sequence database) used for the MaxQuant search.

For phosphorylation:

artmsProtein2SiteConversion(
  evidence_file = "/path/to/the/evidence.txt", 
  ref_proteome_file = "/path/to/the/reference_proteome.fasta", 
  output_file = "/path/to/the/output/ph-sites-evidence.txt", 
  mod_type = "PH")

As a result, the IDs in the “Leading razor protein” column will contain site/peptide-specific notation. For example:

Before: P12345 After: P12345_S23_S45

For ubiquitination:

artmsProtein2SiteConversion(
  evidence_file = "/path/to/the/evidence.txt", 
  ref_proteome_file = "/path/to/the/reference_proteome.fasta", 
  output_file = "/path/to/the/output/ub-sites-evidence.txt", 
  mod_type = "UB")

Tip: How to re-annotate all the Protein columns on the same file.

By default, artmsProtein2SiteConversion doesn’t allow to overwrite the evidence.txt file for security reasons (you don’t want to lose the evidence file if something goes wrong). To overwrite the evidence file the argument overwrite_evidence must be turned on (overwrite_evidence = TRUE).

If the column_name argument is not used, artmsProtein2SiteConversion converts the Leading razor protein column, which is used in the quantification step when protein_groups : remove is selected (default). However, if protein_groups : keep is used, artMS will use the Proteins column. To convert the Proteins column to the site/peptide-specific notation, then add the argument column_name = "Proteins".

To annotate both columns of the same file, first generate the “site-evidence.txt” file, and then use this same output file as the evidence_file and activate overwrite.evidence = TRUE.

In summary, to annotate both the “Leading razor protein” and Proteins columns follow these steps:

# Convert 'Leading razor protein'
artmsProtein2SiteConversion(
  evidence_file = "/path/to/the/evidence.txt", # ORIGINAL
  column_name = "Leading razor protein",
  ref_proteome_file = "/path/to/the/reference_proteome.fasta", 
  output_file = "/path/to/the/phsites-evidence.txt", # SITES VERSION
  mod_type = "PH")

# Convert 'Proteins' 
artmsProtein2SiteConversion(
  evidence_file = "/path/to/the/phsites-evidence.txt", # <- USE SITES VERSION
  column_name = "Proteins",
  overwrite_evidence = TRUE, # <--- TURN ON
  ref_proteome_file = "/path/to/the/reference_proteome.fasta", 
  output_file = "/path/to/the/phsites-evidence.txt", # <- SITES VERSION
  mod_type = "PH")
  1. Generate a new configuration file (phsites_config.yaml or ubsites_config.yaml) as explained above, but using the “new” sites-evidence.txt file instead of the original evidence.txt file:
files:
  evidence : /path/to/the/evidence-site.txt
  keys : /path/to/the/keys.txt
  contrasts : /path/to/the/contrast.txt
  output : /path/to/the/output/results_ptmSITES/sites-results.txt # <- this one
  .
  .
  .
data:
  .
  .
  .
  filters:
    modifications : PH # <- Don't forget this one.

Once the new yaml file has been created, execute:

artmsQuantification(
  yaml_config_file = '/path/to/config/file/phsites_config.yaml')

4.4 Output files

The files generated after succesfully running artmsQuantification are (based on MSstats documentation):

4.4.1 TXT (tab delimited) files

  • results.txt: the results of the quantification in long format, i.e., each line is a protein for each comparison. It includes the following columns:
    • Protein: Protein ID
    • Label: comparison (from contrast.txt)
    • log2FC: log2 fold change
    • SE: standard error
    • Tvalue: test statistic of the Student test
    • DF: degree of freedom of the Student test
    • pvalue: raw p-values
    • adj.pvalue: p-values adjusted among all the proteins in the specific comparison using the approach by Benjamini and Hochberg
    • issue: shows if there is any issue for inference in corresponding protein and comparison, for example, OneConditionMissing or CompleteMissing.
    • MissingPercentage: percentage of random and censored missing in the corresponding run and protein out of the total number of feature in the corresponding protein.
    • ImputationPercentage: percentage of imputation
    • Note: If a protein is completely missed in one of the conditions of a given comparison, it would get the flag OneConditionMission with adj.pvalue=0 and log2FC=Inf or -Inf even though pvalue=NA. For example, if for the comparison Condition A - Condition B one protein is completely missed for condition B, then log2FC = Inf and adj.pvalue = 0. SE, Tvalue, and pvalue will all be NA.
  • results-annotated.txt: same as results.txt but 3 more columns of annotations, i.e., Gene, ProteinName, and EntrezID
  • Different outputs from the MSstats dataProcess step (Normalizing and summarizing data, check MSstats documentation to find out more), including:
    • results_ModelQC.txt
    • results_RunlevelData.txt
    • results-mss-groupQuant.txt
    • results-mss-normalized.txt
    • results-mss-sampleQuant.txt
  • Sample size and power calculation:
    • results_sampleSize.txt
    • results_experimentPower.txt

4.4.2 Plots (pdf)

  • results-heatmap.pdf
  • results-peptidecounts-perBait.pdf
  • results-peptidecounts.pdf
  • results-sign.pdf
  • results-volcano.pdf

5 ANALYSIS OF QUANTIFICATIONS

Comprehensive analysis of the quantifications outputs obtained from the function artmsQuantification() section to find out more). It includes:

  • Annotations
  • Summary files in different formats (xls, txt) and shapes (long, wide)
  • Numerous summary plots
  • Enrichment analysis using Gprofiler
  • PCA of protein abundance
  • PCA of quantifications
  • Clustering analysis

5.1 Inputs

It takes as input two files generated from the previous quantification step (artmsQuantification())

  • -results.txt : MSstats quantification results
  • -results_ModelQC.txt : MSstats normalized abundance values. It will be used to extract details about reproducibility.

To run this analysis:

  1. Set as the working directory the folder with the results obtained from artmsQuantification().
setwd('~/path/to/the/results_quantification/')

And then run the following function (e.g., for a protein abundance “AB” experiment)

artmsAnalysisQuantifications(log2fc_file = "ab-results.txt",
                              modelqc_file = "ab-results_ModelQC.txt",
                              species = "human",
                              output_dir = "AnalysisQuantifications")

A few comments on the available options for artmsAnalysisQuantifications:

  • isPTM: for protein abundance (AB), Affinity Purification-Mass Spectrometry (APMS), and global analysis of posttranslational modifications (PH and UB) use the option "noptm". For a site specific PTM analysis use "ptmsites".
  • species: this downstream analysis supports (for now) "human" and "mouse"
  • enrich: If TRUE, it will perform enrichment analysis using gProfileR
  • isBackground. If enrich = TRUE, the user can provide a background gene list (add the file path as well)
  • mnbr: Minimal Number of Biological Replicates for imputation. Missing values will be imputed and this argument is set to specify the minimal number of biological replicates that are required in at least one of the conditions, but for all the proteins. For example, mnbr = 2 would indicate that only proteins found in at least two biological replicates will be imputed. In addition, mnbr = 2 would also add the constrain that any protein must be identified in at least two biological replicates of the same condition or it will be filtered out. That is, if mnbr = 2, a protein found in two conditions but only in one biological replicate in each of them, it would be removed.
  • l2fc_thres: log2fc cutoff for enrichment analysis, absolute value, e.g., if it is set to 1, it will consider significant log2fc > +1 and log2fc < -1.
  • ipval: select whether pvalue or adjpvalue id used for the analysis. The default option is adjpvalue (multiple testing correction). But if the number of biological replicates for a given experiment is too low (for example n = 2), then pvalue is recommended.

5.2 Outputs

Summary file

Reminder: for any given relative quantification, as for example WT-Mutant:

  • Proteins with positive log2fc values (i.e. log2fc > 0) are more abundant in the condition on the left (WT)
  • Proteins with negative log2fc values (i.e. log2fc < 0) are more abundant in the condition on the right (Mutant)*

The summary excel file (results-summary.xlsx) gathers several tabs:

  • log2fcImputed: includes quantitative results. The columns: self-defining: Protein, Gene, ProteinName, EntrezID, Comparison, log2FC, pvalue, adj.pvalue, imputed. Not self-defining columns: iLog2FC, iPvalue: in addition to the model based log2fc and pvalues, these columns include imputed log2fc and pvalues (when completely missed in one of the conditions), followed by as many columns as the number of conditions, indicating the number of biological replicates in which each protein/ptmsite was identified. CMA stands for “Condition Most Abundant”: condition in which each protein was found more abundantly (based on the log2fc).
  • wide_iLog2fc: log2fc values (including imputed values) in wide format, i.e., each row is a unique protein/ptmsite. The columns shows the values for each of the comparisons.
  • wide_iPvalue: same as before, but for pvalues (including imputed)
  • enrichALL: enrichment analysis using GProfileR for all the proteins changing significantly in any direction (ab(log2fc) > 0 and pvalue < 0.05)
  • enrich-MACpos: enrichment of only the positive significant changes (log2fc > 1, pvalue < 0.05)
  • enirch-MACneg: enrichment of only the negative significant changes (log2fc < -1, pvalue < 0.05)
  • enMACallCorum, enMACposCorum, enMACnegCorum: same as above but only for protein complex enrichment analysis (based on CORUM)

Text files

  • results-log2fc-long.txt: same as the log2fcImputed tab from the summary file
  • results-log2fc-wide.txt: wide version (i.e., each row is an individual protein) of pvalues and adj.pvalues for each comparison

Gene Enrichment analysis: enrichment analysis only supported for human and mouse. Check the GprofileR documentation to find out more about the details:

  • results-enrich-MAC-allsignificants.txt: all significant changes (abs(log2fc) > 1 & pvalue < 0.05)
  • results-enrich-MAC-positives.txt: only positive significant changes (log2fc > 1 & pvalue < 0.05)
  • results-enrich-MAC-negatives.txt: all significant changes (based on p-value only)

Protein Complex Enrichment analysis (based on CORUM)

  • results-enrich-MAC-allsignificants-corum.txt
  • results-enrich-MAC-positives-corum.txt

  • results-enrich-MAC-negatives-corum.txt

  • results-enrich-MAC-allsignificants-corum.pdf
  • results-enrich-MAC-negatives-corum.pdf

  • results-enrich-MAC-positives-corum.pdf

Clustering

  • results.clustering.log2fc.all-overview.pdf
  • results.clustering.log2fc.all-zoom.pdf
  • results.clustering.log2fcSign.all-overview.pdf

  • results.clustering.log2fcSign.all-zoom.pdf

  • results.log2fc-clusterheatmap-enriched.txt
  • results.log2fc-clusterheatmap.txt
  • results.log2fc-clusterheatmap.pdf

  • results.log2fc-clusters.pdf

Correlations

  • results.correlationConditions.pdf
  • results.reproducibilityAbundance.pdf
  • results.correlationQuantifications.pdf
  • results.log2fc-corr.pdf

Miscellaneous

  • results.relativeABUNDANCE.pdf

  • results.distributions.pdf
  • results.distributionsFil.pdf
  • results.imputation.pdf

  • results.TotalQuantifications.pdf

PCA

Based on relative abundance

  • results-pca-pca01.pdf
  • results-pca-pca02.pdf
  • results-pca-pca03.pdf

Based on significant changes

  • results.log2fc-dendro.pdf
  • results.log2fc-individuals-pca.pdf

6 MISCELLANEOUS FUNCTIONS

artMS also provides a number of very handy functions.

6.1 Annotate data.frame with Gene Symbol, Name, ENTREZ based on Uniprot IDs

Takes the given columnid (of Uniprot IDs) from the input data.frame, and map the gene symbol, name, and entre id (source: bioconductor annotation packages)

# This example adds annotations to the evidence file available in
# artMS, based on the column 'Proteins'.

evidence_anno <- artmsAnnotationUniprot(x = artms_data_ph_evidence,
                                        columnid = 'Proteins',
                                        species = 'human')

6.2 Average Intensity, RT, CR

Taking as input the evidence file, it will summarize and return back the average intensity, average retention time, and the average calibrated retention time for each protein. If a list of proteins is provided, then only those proteins will be summarized and returned. Check ?artmsAvgIntensityRT() to find out more options.

artmsAvgIntensityRT(evidence_file = '/path/to/the/evidence.txt)

6.3 Change column name

Changes a given column name in the input data.frame

artms_data_ph_evidence <- artmsChangeColumnName(
                               dataset = artms_data_ph_evidence,
                               oldname = "Phospho..STY.",
                               newname = "PH_STY")

6.4 Individual abundance dot plots

Protein abundance dot plots for each unique uniprot id. It can take a long time

artmsDataPlots(input_file = "results/ab-results-mss-normalized.txt",
               output_file = "results/ab-results-mss-normalized.pdf")

6.5 Enrichment analysis function

Enrichment analysis based on a data.frame with Gene and Comparison/Label protein (i.e, typical MSstats results)

# The data must be annotated (Protein and Gene columns)
data_annotated <- artmsAnnotationUniprot(
                      x = artms_data_ph_msstats_results,
                      columnid = "Protein",
                      species = "human")
# And then the enrichment
enrich_set <- artmsEnrichLog2fc(
                   dataset = data_annotated,
                   species = "human",
                   background = unique(data_annotated$Gene), 
                   verbose = FALSE)

6.6 Enrichment analysis using gProfileR

Function that simplifies enrichment analysis using gProfileR

# annotate the MSstats results to get the Gene name
data_annotated <- artmsAnnotationUniprot(
                                     x = artms_data_ph_msstats_results,
                                     columnid = "Protein",
                                     species = "human")

# Filter the list of genes with a log2fc > 2
filtered_data <- 
unique(data_annotated$Gene[which(data_annotated$log2FC > 2)])

# And perform enrichment analysis
data_annotated_enrich <- artmsEnrichProfiler(
                                   x = filtered_data,
                                   categorySource = c('KEGG'),
                                   species = "hsapiens",
                                   background = unique(data_annotated$Gene))

6.7 MaxQuant evidence file to SAINTexpress format

Converts the MaxQuant evidence file to the 3 required files by SAINTexpress. Choose one of the following quantitative MS metrics:

  • MS spectral counts (use msspc)
  • MS intensities (use msint)
artmsEvidenceToSaintExpress(evidence_file = "/path/to/evidence.txt", 
                            keys_file = "/path/to/keys.txt", 
                            ref_proteome_file = "/path/to/org.proteome.fasta")

6.8 MaxQuant evidence file to SAINTq format

Converts the MaxQuant evidence file to the required files by SAINTq. The user can filter based on either peptides with spectral counts (use msspc) or all the peptides (use all) for the analysis. The quantitative metric can be also chosen (either MS intensity or spectral counts)

artmsEvidenceToSAINTq(evidence_file = "/path/to/evidence.txt", 
                      keys_file = "/path/to/keys.txt", 
                      output_dir = "saintq_input_files")

6.9 Generate Phosfate input file

It generates the Phosfate input file from the imputedL2fcExtended.txt file resulting from running the artmsAnalysisQuantifications() on a ph-site quantification (see above). Notice that the only species suported by PHOTON is humans.

artmsPhosfateOutput(inputFile = "your-imputedL2fcExtended.txt")

6.10 Generate Photon input file

It generates the Photon input file from the imputedL2fcExtended.txt file resulting from running the artmsAnalysisQuantifications() on a ph-site quantification (see above). Please, notice that the only species suported by PHOTON is humans.

artmsPhotonOutput(inputFile = "your-imputedL2fcExtended.txt")

6.11 Remove contaminants and empty proteins from the MaxQuant evidence file

Remove contaminants and erroneously identified ‘reverse’ sequences by MaxQuant, in addition to empty protein ids

evidencefiltered <- artmsFilterEvidenceContaminants(x = artms_data_ph_evidence)

6.12 Generate ph-site specific evidence file

Generate extended detailed ph-site file, where every line is a ph site instead of a peptide. Therefore, if one peptide has multiple ph sites it will be breaking down in multiple extra lines for each of the sites.

artmsGeneratePhSiteExtended(df = dfobject, 
                            species = "mouse", 
                            ptmType = "ptmsites",
                            output_name = log2fc_file)

7 METABOLOMICS

artMS enables the relative quantification of untargeted polar metabolites using the alignment table generated by MarkerView. This means that the metabolites do not need to have an id in order to perform the quantification, as the m/z and retention time will be used as identifiers.

MarkerView is an ABSciex software that supports the files generated by Analyst software (.wiff) used to run our specific mass spectrometer (ABSciex Triple TOF 5600+). It also supports .t2d files generated by the Applied Biosystems 4700/4800 MALDI-TOF.

Markview is used to align mass spectrometry data from several samples for comparison. Using the import feature in the software, .wiff files (also .t2d MALDI-TOF files and tab-delimited .txt mass spectra data in mass-intensity format) are loaded for retention time alignment. Once the data files are selected, a series of windows will appear wherein peak finding, alignment, and filtering options can be entered and selected. These options include minimum spectral peak width, minimum retention time peak width, retention time and mass tolerance, and the ability to filter out peaks that do not appear in more than a user selected number of samples.

The alignment file is further processed and formatted to perform QC and relative quantification using the following artMS functions:

7.1 Convert Metabolomics

Pre-process the markview .txt file to generate an “evidence-like” file by running:

artmsConvertMetabolomics(input_file = "markview-output.txt", 
                         out_file = "metabolomics-evidence.txt")

7.2 QC Metabolomics

Perform quality control analysis on the metabolomics data by running:

artmsQualityControlMetabolomics(evidence_file = "metabolomics-evidence.txt",
                                keys_file = "metabolomics-keys.txt")

It generates the following plots:

  • plotINTDIST.pdf contains both Box-dot plot and Jitter plot of biological replicates based on MS (raw) intensity values.
  • plotREPRO.pdf correlation dotplot for all the combinations of biological replicates of conditions, based on MS Intensity values using features (mz_rt+charge)
  • plotCORMAT.pdf, includes up to 3 pdf files for technical replicates, biological replicates, and conditions. Each pdf file contains:
    • Correlation matrix for all the biological replicates using MS Intensity values,
    • Clustering matrix of the MS Intensities and correlation distribution
    • histogram of the distribution of correlations
  • plotINTMISC.pdf the pdf contains several pages, including bar plots of Total Sum of Intensities in BioReplicates, Total Sum of Intensities in Conditions, Total Feature Counts in BioReplicates, Total Feature Counts in conditions separated by categories (INT: has a intensity value NOINT: no intensity value ) Box plots of MS Intensity values per biological replicates and conditions; bar plots of total intensity by bioreplicates and conditions; Barplots of total feature counts by bioreplicates and conditions.

7.3 Relative Quantification:

The relative quantification is performed using MSstats. It requires a configuration file (yaml format, please check above). A template can be generated by running: artmsWriteConfigYamlFile(config_file_name = "metab_config.yaml"). The relative quantification is performed by running:

artmsQuantification(yaml_config_file = "metabConfig.yaml")

8 TESTING FILES

The artMS package provides the following testing datasets

Phosphoproteomics dataset: example dataset consisting of two head and neck cancer cell lines (conditions "Cal33" and "HSC6"), 2 biological replicates each). The number of peptides was reduced to 1/8 due to bioconductor limitations on data size.

  • Evidence file: artms_data_ph_evidence
  • Keys file: artms_data_ph_keys
  • Results file: artms_data_ph_msstats_results: results after running artmsQuantification() on the reduced version

The full data set (2 conditions, 4 biological replicates) can be found at the following urls:

  • url_evidence <- 'http://kroganlab.ucsf.edu/artms/ph/evidence.txt'
  • url_keys <- 'http://kroganlab.ucsf.edu/artms/ph/keys.txt'

Protein Complexes dataset: downloaded (2017-08-01) from CORUM database
and further enriched with annotations of mouse mitochondrial complexes not available at CORUM. Used for complex enrichment calculations.

  • artms_data_corum_mito_database

Pathogens Uniprot IDs:

  • artms_data_pathogen_LPN: Legionella pneumophila philadelphia (downloaded 2017-07-17)
  • artms_data_pathogen_TB: Mycobacterium tuberculosis strain ATCC 35801 / TMC 107 / Erdman (downloaded 2018-04-01)

Check the individual help pages (e.g, ?artms_data_ph_evidence) to find out more about them.

9 HELP