ProteinPilot™ Report for ProteinPilot™ Software
Detailed analysis of protein identification / quantitation results automatically
Sean L Seymour, Christie Hunter
Powerful mass spectrometers like the TripleTOF® 6600 and 5600 Systems can rapidly generate extremely large amounts of data. For today’s researchers, tools that can logically and efficiently distill the massive amounts of data down into easily interpretable results are critical. ProteinPilot Software is a powerful, robust, easy to use software tool for protein identification and quantification for discovery research and protein characterization1. With its hybrid sequence tag and database search approach using feature probabilities, the powerful Paragon™ Algorithm can search for hundreds of modifications and sequence variants in a single search2. Coupled with the Pro Group™ Algorithm for protein inference analysis, peptide results are condensed down to the most defensible set of detected proteins with ambiguity among multiple accession numbers reported when appropriate.
In addition to identification and quantitation information, there are many different types of post-acquisition analysis that can be performed that are highly valuable to the protein researcher to ensure results quality and enable workflow refinement. Many of these types of analysis have been combined into a single Excel-based processing tool, the ProteinPilot Report.
Basic reporting for ease of publication
For every ProteinPilot Software database search, a detailed false discovery rate (FDR) analysis is performed and a rigorous report is generated, detailing the quality of protein and peptide identifications3. FDR analysis is performed at the spectral, peptide and protein level (Figure 1). A novel non-linear fitting method is used to determine both a global and a local FDR from the decoy database search3.
A detailed meta-data report is generated which contains a large amount of information that is useful for reporting search details for publication.
Characterization of MS acquisition
One of the keys to fully optimizing the quality of data acquired by an LC-MS system is the ability to measure the appropriate quantitative metrics on the acquisition. The ProteinPilot Report provides many helpful metrics on data quality. For MS data quality, detailed analysis of mass accuracy is performed, both overall (Figure 2, top) and as a function of retention time or precursor signal. Distributions of the charge state, mass, and m/z of confidently identified peptides are generated (Figure 2, middle). Using the precursor intensity at the peak apex, many different valuable analyses are performed, such as the precursor distribution (Figure 2, bottom) which directly measures the dynamic range of detected peptides in a sample.
Characterization of LC properties
Another key aspect to high quality LC-MS analysis is the quality of the chromatography. A dashboard showing all the key properties of chromatography is available. A detailed analysis is performed on the LC peak width for each peptide and plots showing the distribution of peak widths and the median peak width as a function of retention time (Figure 3). This information can be used to assess and improve the LC separation and also during method optimization for quantitative workflows such as SWATH Acquisition or MRMHR workflow. An analysis is also performed to understand how where the MS/MS is triggered relative to the LC peak apex.
Characterization of sample properties
Proteases do not have perfect cleavage specificity. Thus, the ability of the Paragon Algorithm to search for missed cleavages (under cleavage) and unexpected cleavages (over cleavage), in addition to hundreds of sample preparation and biological modifications ensures higher fidelity in the identification results. The Report provides a detailed analysis of the quality of the digestion (Figure 4). Monitoring the missed cleavage and semi-tryptic rates observed in each study is an effective way to ensure that the digestions are working well and reproducibly (Figure 4, top). The heat map (Figure 4, bottom) shows the cleavage rates observed between each residue pair for the cases where digestion did not conform to expected digestion sites.
The Paragon Algorithm in Thorough mode automatically searches for 100s of sample preparation and biological modifications as well as amino acid substitutions. A detailed summary is provided as well as a distillation of the 25 most frequent modifications observed in the confidently identified peptides (Figure 5). It also computes the fraction of total ion signal having the modification of all forms of the same base sequences, as measured via peptide elution apex intensities. This allows for the rigorous QC of sample preparation steps, like cysteine alkylation, and labeling chemistries as well as undesired side reactions.
Characterization of Quantitative Results
There are a number of dashboards that are computed to help with understanding the quality of the quantitative data obtained for the SCIEX iTRAQ® reagents or other labeling experiments analyzed. One important analysis that is done on a quantitative dataset is a target-decoy analysis of the quantitative ratios to determine the p-value cutoff to use to get a desired FDR level in the differential protein list. This can be done when there is a true analytical replicate present in the multiplex that can be used to create decoy ratios (Figure 6). Once the p-value is determined the final list of proteins can be easily pull from the tab that distills the list of differentially expressed proteins sorted by ascending p-value.
Visualization of individual protein results is possible using the protein viewer (Figure 7). Here the underlying quantitative data for specific proteins can be visualized.
After every ProteinPilot Software database search (when searching the UniProt/SwissProt FASTA files), the UniProt website is accessed and the ontology information available for every identified protein is downloaded and incorporated into the results (*.group file). The report performs an analysis on this information and determines if there is any enrichment of any of the protein classes in the dataset or specifically in the differentially expressed proteins (Figure 8).