APOSTL is an interactive affinity proteomics analysis software developed to reformat affinity proteomics data (both spectral counting and MS1) for input into the SAINTexpress statistical package\(^1\) and to visualize the output(s). APOSTL was developed at H. Lee Moffitt Cancer Center & Research Institute and distributed under a GNU General Public License (GPL). APOSTL is built in Python and R and integrated with SAINTexpress into a cohesive affinity proteomics data analysis package using the Galaxy framework.
Within the Galaxy environment APOSTL is able to recognize either a Scaffold (Proteome Software) Samples Report file or the peptides.txt file in the MaxQuant (link) txt output folder. No modifications should be made to these files. Using the Bait Create tool, you can create your bait.txt file. It is important that the individual bait names match the bait names within your Scaffold or MaxQuant output. APOSTL uses the bait file to find the user's baits of interest. Additionally there is an option to make the prey file (Y/N).
When making a prey file, APOSTL queries a user provided FASTA database in order to extract protein amino acid lengths and gene names. This may take several minutes depending on your computer and if your Galaxy distribution is cluster enabled. Some users may want to run SAINTexpress using the same data set while changing which baits are considered test or control. It is useful to toggle Make Prey off in order to save time by circumventing this step as the same prey file can be used for both SAINTexpress runs.
Supported Inputs:
Note: All files must be tab delimited txt
Once SAINTexpress has been run, APOSTL is able to read the resulting list.txt file. From here APOSTL calculates normalized spectral abundance factor\(^2\) (NSAF) values for each prey based on the average spectra observed for each bait. Additionally, APOSTL will calculate the NSAF Score as described below. Optionally, APOSTL calculates the probability of a specific interaction based on prey prevalence in the CRAPome.
Within an interactive analysis environment, APOSTL offers various visualization and analysis options including:
1) Correlations between replicate runs
2) Boxplots of selected protein across all replicates
3) Bubble graphs for each bait with numerous customization options:
\(SpC\) = Spectral Count (or MS1 Intensity)
\(L\) = Amino Acid Length
\(FC\) = Fold Change
\(ε = 1/mean(\frac{SpC}{L})_c\)
\(n\) = # control replicates
4) Density plots of all baits for options specified above
5) Cytoscape networks of all preys passing filtering criteria displaying newly identified bait-prey interactions
6) Filtered data displayed as a table
7) KEGG pathway analysis of filtered data
8) Gene Ontology (GO) term analysis of filtered data
All graphs, images and data can be saved using the Download buttons or alternatively by RMB + Save As to generate 600 dpi figures.
Within the Galaxy environment, APOSTL offers several other analysis options including:
1) Protein-protein interaction networks querying known PPI as annotated in ConsensusPathDB following filtering for simple cytoscape import using the import network from file option.
2) Dot plot and clustering analysis from the the Prohits Visualization Tool\(^3\)
Teo G, Liu G, Zhang J, Nesvizhskii AI, Gingras A-C, Choi H. SAINTexpress: improvements and additional features in Significance Analysis of INTeractome software. J Proteomics. 2014 Apr 4
Zybailov B, Mosley AL, Sardiu ME, Coleman MK, Florens L, Washburn MP. Statistical Analysis of Membrane Proteome Expression Changes in Saccharomyces cerevisiae. J Proteome Res. 2006 Apr 10
Knight JD, Liu G, Zhang JP, Pasculescu A, Choi H, Gingras AC. A web-tool for visualizing quantitative protein-protein interaction data. Proteomics. 2015 Jan 19
APOSTL support is provided by the Haura and Rix labs:
Source code is available on Github
Below are some occasional errors that we have encountered while testing APOSTL with solutions:
Error: length of 'dimnames' [2] not equal to array extent This error is usually due to the user filtering data such that it has removed all the points in your data. Adjust your filtering to resolve.
Occasionally when rendering the Cytoscape network in the browser, node shapes can become white when changing them. This issue has not been fully resolved in the Beta Release of this tool. To fix this issue currently, toggle different shapes in your network until the shape you want remains.
Error: invalid request, server returned Success: (200) OK (http://rest.kegg.jp/link/hsa/pathway) This is usually due to your R Bioconductor installation being out of date. It can usually be resolved by running the BiocInstaller::biocLite() command in R on the machine hosting the Shiny server and Galaxy.
BETA Release: The network refreshes itself and DOES NOT save user layouts when altering parameters. Make sure to finalize filtering criteria and aesthetics before reorganizing the network. There remains some instability in the node shapes with a large number of nodes present. Try multiple shapes to resolve the issue.