Workflow

Data import

First, the data needs to be imported.

Currently the mspypeline package supports the analysis of label-free shotgun proteomics analyzed by the MaxQuant software. For a complete analysis, the mspypeline package deploys several MaxQuant output tables, however, the minimal requirement to perform exploratory analysis is only the proteinGroups.txt file that contains aggregated protein intensities.

Since several different mass spectrometry techniques exist and other software programs like MaxQuant are applied a number of different file formats containing mass spectrometry data exist. So far, analysis of files with other formats is not supported as mspypeline requires a strictly followed internal data format.

For this purpose, FileReaders are involved that translate the particular format of each file into the internal file format. Until now, only a MQReader is provided.

The MQReader can reformat several output tables provided by MaxQuant including the following text files:

• proteinGroups

• peptides

• summary

• parameters

• msScans

• msmsScans

• evidence

If present, these files are utilized to generate the quality control report.

Upon data import the data gets internally formatted and prepared for the analysis. Default preparation of the data by the MQReader includes:

• the removal of “Reverse” proteins, those “Only identified by site” or marked as “Potential contaminant”

• proteins that are missing both an identified gene name and a FASTA header are discarded

• intensities of proteins assigned with an identical gene name (duplicates) are handled (sum or drop).

These default configurations can be customized by the user in accordance with their preferences.

During the attempt to import the data, a first quality test is performed, which verifies whether all required files and directories are provided and if they are correctly structured.

Tip

• To perform data analysis with different intensity types (e.g. LFQ or iBAQ) it is necessary to specify these options for the MaxQuant analysis.

• It is recommended to select the information of FASTA file headers for the detected proteins as mspypeline will use the gene name annotation therefrom to index the detected proteins.

• To ensure proper analysis the samples of the experiment have to be named according to the naming convention.

Warning

• the output folder, called txt, and all contained files from a MaxQuant run must not be renamed or the analysis will not work.

Quality control

Following data import it is recommended to first prepare a control report on the raw data in order to obtain a preliminary impression of the data quality. This quality control report is specifically designed to process the information from the distinct MaxQuant files (if available) to generate a multi-page pdf document. Here, the quality of the raw data can be investigated of both experimental and technical MS specific parameter.

Such a quality control is provided by the MaxQuant plotter. An exemplary MaxQuant report is provided in the gallery.

Data Preprocessing

Data may be processed in multiple ways and this can substantially alter the results of your analysis.

Data preprocessing available in mspypeline comprises: * the choice of protein intensities provided by MaxQuant: raw, label-free quantification (LFQ) or intensity-based absolute quantification (iBAQ) intensities * averaging technical replicates * removal of erroneous samples * normalization and standardization of the data set.

Intensity options

• LFQ Intensity (“lfq_log2”) (extended information)

• raw Intensity (“raw_log2”)

• iBAQ Intensity (“ibaq_log2”) (extended information)

Regardless of the choice of protein intensity, the GUI handles all data in log2 format. However, it is possible to analyze the data without log2 scale (“lfq”, “raw”, “ibaq”) if advanced data analysis is performed by interacting with the package programmatically.

Normalization options

• No normalization

• Median Normalization via: MedianNormalizer

• Quantile Normalization with missing value handling via: QuantileNormalizer and interpolate_data()

• Tail Robust Quantile Normalization (TRQN) via: TailRobustNormalizer and QuantileNormalizer

• TRQN with missing value handling via: same as above and interpolate_data()

• Tail Robust Median Normalization via: TailRobustNormalizer and MedianNormalizer

To aid the determination of the best possible normalization method, two plots may be created: plot_normalization_overview() and plot_heatmap_overview_all_normalizers().

These methods will output a multipage PDF file in which the data is plotted repeatedly after applying the different normalization options. Thereby it is possible to get a better understanding of the effect of each normalization method on the data.

Please read the function description explaining how normalized data should look like. Once a normalization method is chosen, it is highly recommended to perform all further analysis with the same normalized data.

Exploratory Analysis

Create outlier detection and comparison plots

The descriptive and comparison plots can for example help to analyze how biological replicates compare to each other or how different conditions effect detected proteins.

Create statistical inference plots

Statistical inference plots can inform about differential protein intensities between groups of the data set. The calculation of statistical significances of the variation of protein intensities between groups can help to exploit biological questions by incorporation the functional profile of proteins or protein sets.

Statistics for each plot are calculated based on the intended usage of the plot.

• for the plot_pathway_analysis() an independent t-test is applied

• for the plot_go_analysis() a fisher’S exact test is applied

• for the plot_r_volcano() plot the moderated t-statistics is applied which is implemented by the R package limma. Additional R packages might be downloaded when this plot is created for the first time.

Select pathways and GO-Terms of interest

Select Pathways. Selected pathways has following effects:

• for the plot_pathway_analysis() one plot per pathway is created

• in the plot_rank(), if a protein is found it is marked on the plot and colored by the pathway

• in the plot_r_volcano(), if a pathway is selected, proteins of that pathway are annotated in the plot instead of the most significant proteins that are annotated by default

Select GO Terms. Selected GO-Terms has following effects:

• for the plot_go_analysis() one additional barplot is added per GO term