How LOD Proteomics Accelerates Deep Research with Lower Risk

Was this article helpful?

It’s easy to snorkel in the shallows among tons of colorful fish. But if you want to do marine science, you start with quality scuba gear and go deep with a dive-master. Deep diving with DIY equipment and know-how can be painful or worse. 

Proteomics was once the “Next Big Thing” to revolutionize clinical discovery but fell short. We discovered why — the field is stuck in the shallows with “fast-and-cheap” statistical data analysis. 

We recently invented LOD (limit-of-detection) proteomics to identify/characterize protein forms down to LOD of mass spectrometers. This conceptually simple paradigm requires specialized equipment (data platform) and know-how to mine >98% of detected MS1 peptides inaccessible by today’s statistical workflows.

Current research can be like shooting in the dark. To study PTMs, researchers try a few DIA m/z ranges for MS2 fragmentation hoping to find some peptides of interest. This traditional approach is costly in time and resource, cannot work for low-abundance peptides, and has a high chance of failure. 

LOD proteomics uses a parallel digital track to enhance and accelerate existing wet lab research with lower risk. Unlike current MS2 software limited to the <2% tip of the iceberg, it builds a MS1 2D “map” from every MS run to optimize the next DIA run. It lets you know exactly what to expect and why before every DIA run. 

Since LOD proteomics can be piggybacked on to any existing research project as a MS1 cheat-sheet, it eliminates concern about peer-review. 

There are two ways to accelerate your research with LOD proteomics. Labs can buy a SORCERER™ AIMS system (physical or cloud-based) or contract us as a service. Either way, there will be custom services for initial training and custom scripts. 

LOD proteomics from first principles

The basic concept was recently published [see poster here] and presented to top researchers with zero pushback. I believe all were surprised at its elegant simplicity. But it only works where there is enough good accurate data. 

In high-accuracy MS, it’s useful to view each raw data-point as one detected raw ion with (usually) accurate m/z and rough intensity. The significance: each becomes standalone evidence down to LOD; there is no need to use any math or fit a curve. 

Here, data analysis is simply a “cyber” assay where we gather raw ions that support the identification and gross quantitation for specific peptides of interest. Overwhelming evidence is convincing; too little evidence is not; and something in-between requires expert judgment, period. No need for complex statistics. 

In MS2 data, +1 y-ions can be both ever-present and accurate (typ. <0.003 tolerance), unlike b-ions or higher charges. This means matching almost all low +1 y-ions uniquely defines the c-terminus sequence including any modifications. (DIA just means multiplexed spectra with multiple y-ion chains.) We find that 8 near-consecutive amino acids are usually enough to uniquely identify the protein, sometimes more and sometimes less. (Just use ‘grep’ to check.)

For each MS2 peptide ID, we select two representative MS1 ions — the apex ion and its isotopic twin — to calculate its charge, accurate mass, and gross quant.

For a LOD peptide with a pair of MS1 isotopic ions, we can also calculate mass and gross quant, but with anonymous identity. Our poster showed it is possible to infer identity by MS1 triangulation. For example, we can infer the identity of a LOD peptide whose mass is exactly one phosphorylation less than a singly-phosphorylated peptide ID. 

Putting it all together, it requires about 12 raw ions (4 MS1, 8+ MS2) to infer the identity/quantitation of an anonymous MS1 peptide by referencing its mass difference from identified peptides. 

Re-introducing Sage-N Research

For the record, we’re not really a software company per se, but a technology IP company. As a Silicon Valley system engineer from MIT with no biology background, I saw proteomics as my opportunity to give back to the American Dream. In a way, I’ve already contributed: my patent 5483478 [click here] sped up all FPGA chips that accelerate today’s mass spectrometers. 

When our company was founded, my plan was to sell enough to learn the field then (quickly) invent/patent technologies that change the world. Better late than never, LOD proteomics is poised to redefine ultra-sensitive protein research. 

We look forward to helping the next generation of pioneers to mine 100x more hidden peptides for world-changing discoveries.

For more information, please contact Ms. Terri Nowak [TNowak@SageNResearch.com].