edwardmeng132 / prince-matlab

Bioinformatics pipeline for predicting protein interactomes via co-elution

PrInCE

Predicting Interactomes via Co-Elution: a bioinformatics pipeline for analyzing PCP-SILAC and other co-elution experiments.

• Predict protein-protein interactions
• Predict protein complexes
• Calculate protein abundance differences

See FAQS and TablesAndFigures for detailed explanation of output files.

Original versions of the analysis were written by Anders Kristensen¹ and Nichollas Scott².

How to use

1. Download pipeline.
2. Format your data.
3. Put your data in the input folder.
4. Run prince.m.

1. Download PrInCE.

Click the green "Clone or Download" button in the top right of the github page (https://github.com/fosterlab/PrInCE-Matlab). Then click "Download ZIP".

2. Format your data.

Data files

Make one csv file for each biological condition. Except for the header, each row is co-elution data from a single protein and replicate, e.g. a chromatogram. File names must be of the format "condition1.csv", where "1" denotes the experimental condition. For example, for an experiment with two conditions, name the data files "condition1.csv" and "condition2.csv". Each file is formatted like this:

• Column 1: Protein ID (must match with reference database, e.g. CORUM)
• Column 2: Replicate number (integer)
• Columns 3-end: Protein amount, e.g. isotopologue ratio for PCP-SILAC

Typical datasets have hundreds or thousands of proteins and dozens of fractions. Here's a simple dataset with one condition, two replicates, and five fractions:

Important: Ensure that files are "saved as csv", e.g. Excel --> "Save as" --> "Save as csv"

Reference database of known protein complexes

PrInCE needs a reference database of known protein complexes, e.g. CORUM. This reference database must be a file in the same format as CORUM's allComplexes.txt file (downloadable here), and we recommend this for mammalian datasets. In case a custom reference must be made, it must follow this format:

• must be tab-delimited
• must have a header
• reference complexes are in the sixth column, and members of a complex are semicolon-separated

Note: protein IDs in data files must match a subset of protein IDs in the reference database.

3. Unzip the pipeline, and put your data in the Input folder.

Unzip the downloaded zip file (PrInCE-master.zip). This creates four folders and a few files, like this:

• PrInCE-master/
  • Code/
  • Input/
    • Reference database file (e.g allComplexes.txt)
    • Data file 1 (e.g condition1.csv)
    • Data file 2 (e.g condition2.csv)
    • ...
  • Output/
  • Standalone/
  • experimental_design.rtf
  • prince.m

Place all data files (there will be one for each biological condition) and the reference database file (typically "allcomplexes.txt" from CORUM) in the Input/ folder.

Important: PrInCE includes test data files (see below). These test files should be replaced/deleted before analyzing your data!

4. Run analysis.

When running PrInCE in Matlab or as a standalone program, you'll first be prompted for input files and experiment parameters.

1. Input files. Major protein groups and reference database. See FAQ for what is the major protein groups file?;
2. Experiment parameters. For experiments with multiple conditions, fold changes are calculated as a ratio of protein amounts between treatment and non-treatment conditions, e.g. condition2/condition1. Treatment is the numerator, non-treatment the denominator. To set treatment and non-treatment, enter integers in the boxes that correspond to condition files (condition1.csv, etc). Interaction precision controls the quality-vs-quantity of the predicted interactions. Typical values are 50%, 60%, and 75%. A 50% precise interaction list has half true positives and half false positives.
3. Skip modules? If unsure, leave this section alone! If PrInCE was previously run and you want to start the analysis from a module (e.g. Interactions), you can skip the preceding modules. Alternatively you can run a partial analysis, e.g. skipping the Complexes module.
4. Skip plots? GaussBuild and FoldChanges make many plots of individual proteins! You can choose to skip these here.
5. Click this to run PrInCE. As the pipeline runs, output figures and tables will be generated in the Output/ folder.

Important: If you don't provide input files or experiment parameters PrInCE will use default values. These defaults work for the test data but probably won't describe your data, so enter your experiment details!

Running PriNCE in Matlab

If you have Matlab R2016a or later, we recommend you run the Matlab source code. In Matlab, navigate to the PrInCE-master folder you made in step 3. In the Matlab command line type

prince.m

Running PriNCE as standalone program

Mac and PC users can run PrInCE without owning Matlab. This requires downloading Matlab Runtime (~1GB) and setting up the appropriate folders, so is not recommended if a copy of Matlab is available. To run a standalone version of PrInCE:

1. Run "PrInCEInstaller_web". _For Mac users, this is Standalone/Mac/PrInCEInstaller_web. PC users can download the standalone version here (file exceeds github's 100MB limit).
2. Complete the installation process.
   • When prompted to choose the first installation folder (prince standalone), choose the PrInCE-master folder.
   • For the second installation folder (Matlab runtime), keep the default.
   • The "Confirmation" screen shows the folders that prince standalone and Matlab runtime are in. Mac users, copy and paste the paths to these TWO folders! You'll need them in step 4 below.
3. Move prince.exe (Windows) or prince.app and run_prince.sh (Mac) to the same folder as prince.m (i.e. the PrInCE-master folder).
   • These files are in the "application" folder created by PrInCEInstaller_web.
4. Run the standalone prince application.
   • Mac: Open a Terminal window. Navigate to the PrInCE-master folder by typing cd /path/to/PrInCE-master/, where "/path/to/PrInCE-master/" is the first path copied from the Confirmation screen (see above). Run prince by typing ./run_prince.sh /path/to/Matlab/Runtime/, where "/path/to/Matlab/Runtime/" is the second path copied from the Confirmation screen. Typical location is /Applications/MATLAB/MATLAB_Runtime/v92/. e.g. in Terminal, after navigating to the PrInCE-master folder, type ./run_prince.sh /Applications/MATLAB/MATLAB_Runtime/v92/.
   • Windows: Run prince.exe (double-click the prince.exe icon). N.B. Only double-click prince.exe once! There's a lag of 5-10 seconds. A common mistake is to start prince.exe multiple times.

Important: Windows users may have to turn off the "check apps and files" in Windows Defender.

Test data

PrInCE comes with test co-elution data from a recently published paper³. These files are "condition1.csv", "condition2.csv", "Major_protein_groups.csv", and "allComplexes.txt" in the Input/ folder. They give examples of what correctly formatted data files look like and ensure that you can run PrInCE correctly. To do the latter, just follow the instructions in step 5! If all is good, the Output/ folder will be populated with figures and data files within a few hours.

Important: Remove/delete these 4 test files before running your own data!

References

1. Kristensen AR, Gpsoner J, Foster LJ. A high-throughput approach for measuring temporal changes in the interactome. Nat Methods. 2012 Sep;9(9):907-9. doi: 10.1038/nmeth.2131.
2. Scott NE, Brown LM, Kristensen AR, Foster LJ. Development of a computational framework for the analysis of protein correlation profiling and spatial proteomics experiments. J Proteomics. 2015 Apr 6;118:112-29. doi: 10.1016/j.jprot.2014.10.024.
3. Scott NE, Rogers LD, Prudova A, Brown NF, Fortelny N, Overall CM, Foster LJ. Interactome disassembly during apoptosis occurs independent of caspase cleavage. Molecular Systems Biology. 2017 13(1), 906. doi: 10.15252/msb.20167067