Systems Approach to Innovation Helps Extract More Data

  • Technology lag affects the study of proteins, which when quantified can indicate disease or wellness.
  • ISB uses systems approach to innovate and transcend technology limitations.
  • ISB co-develops new technique and software suite to increase protein detection rate.

By Dr. Kristian Swearingen

Scientific breakthroughs are often contingent on available technology. In the case of measuring proteins, the mass spectrometers that exist today aren’t fast enough to detect all the proteins that might be present in a given sample. So it’s necessary for researchers to create ways to push the current technology in order to improve protein detection capabilities.

The latest advancement is the result of a collaboration between ISB and a team of scientists at Stockholm University and KTH in Sweden to develop a new software technique for extracting more protein identifications from experimental data. The technique, dubbed iAMT (in silico accurate mass and time), was published in September in the Journal of Proteome Research. ISB researcher Michael Hoopmann in the Moritz group conceived iAMT and co-developed the suite of software tools that drive the method. In addition to enhancing future efforts, iAMT can be used to re-analyze previous experiments to find proteins that were missed.

The study of proteins by the technique called proteomics is important because the quantitative differences of certain proteins can indicate disease or wellness. Genes encode proteins, and there are many different interconnected biological processes taking place at any given time that can influence or be influenced by whether a gene expresses those proteins. The systems biology approach to proteomics is to study the proteome as a dynamic network of interacting components. It is therefore critical to systems biology to be able to detect as many proteins as possible.

In a typical proteomics experiment, proteins are cleaved into fragments called peptides, which are separated by liquid chromatography and analyzed in a mass spectrometer – a technique called “shotgun proteomics.” Within the mass spectrometer the peptides are isolated one at a time and fragmented. The resulting signal is compared against expected signals predicted from the gene database and used to identify the peptide. If enough quality peptide identifications are made, the protein identity can be inferred.

Currently, it is not possible to identify or quantitate every protein present in a complex sample, such as human cells. The human proteome is estimated to consist of more than 600,000 unique peptides with millions more when post-translational modifications are also considered. Detecting each of these peptides in a single experiment would require an instrument many times faster than existing mass spectrometers to individually fragment the hundreds of thousands of peptides present in a sample. Consequently, only a fraction of the peptides in a sample are selected for fragmentation, and of these, an even smaller fraction produce signal of sufficiently high quality to give a confident identification.

ISB researchers and their collaborators developed iAMT’s suite of software tools that can increase the number of protein identifications in complex samples by up to 7 percent by finding evidence for proteins that were not identified by MS/MS fragmentation. Even if a peptide of a particular protein is never fragmented, its mass may still be recorded by the mass spectrometer but this information is not used by current mass spectrometry peptide identification algorithms. This information, combined with the chromatographic retention time, can be used to narrow down the potential identity of the peptide to a handful of candidates. Sophisticated algorithms were developed as part of this collaboration between ISB and KTH that predict the behavior of peptides thought to be present and search the data for evidence of real, unidentified peptides that match the predictions. Rigorous statistical analysis of this indirect evidence increases the percentage of high-confidence protein identifications. Additionally, by detecting more peptides, the method increases the overall confidence level of protein identifications.

While researchers wait for proteomics technology to improve, new ways of analyzing data will enable researchers to push existing technology to its limits. The iAMT approach is an example of using systems level thinking to extract all the information possible from experimental data.