SYSTEM AND METHOD FOR DETERMINING GLYCAN TOPOLOGY USING TANDEM MASS SPECTRA
- 技术优势
- ·It is equippedwith a paradigm shift invention that uses machine learning to learnfragmentation rules/patterns to distinguish different fragmentation ions inmass spec data, which can be used to better rank the topology candidatesinferred from mass spec data· Thecomputational complexity is significantly lower than previously reportedmethods and results in lower and more efficient computation time.
- 详细技术说明
- AMethod for de novo Reconstructing Glycan Structures from Mass Spec Data
- *Abstract
-
Glycosylationis a common modification by which a glycan (or oligosaccharide) is covalentlyattached to a target biomolecule such as protein and lipids. Glycans are treeensembles of monosaccharides links via glycosidic bonds formed by thecondensation reaction between the hemiacetal group of one monosaccharide (thenon-reducing end residue) and a hydroxyl group of another (the reducing endresidue). Glycosylation serves important purposes in many biological processes,including protein folding and clearance, cell adhesion, and immunologicalresponses among others. Additionally, glycosylation is one of the key factorsthat determine the solubility, stability, and efficacy of manybiopharmaceuticals. Change in glycosylation patterns are observes under variousdisease conditions. Glucan structural analysis is essential for understandingtheir diverse roles in biological systems, however it is a challenging task dueto the cast number of topologies that they may assume even for a moderate sizedglucan.
Currently,mass spectrometry has become of the most powerful tools for determining glycanstructures. Several processing tools exist for determining the topologies ofglycans using mass spectrometry. Generally, glycan reconstruction methods use acatalog-library approach that searches experimental mass spectra againstpre-built glycan databases. The accuracy of the search results depends not onlyon the quality of the query (i.e. the tandem MS data) but also on thecompleteness of the databases. Because glycan databases are generallyincomplete, it is necessary to develop a de novo method for determination ofglycan structures from their experimental spectra. With enough information (e.g., precursor ion mass, possible monosaccharide components, charge carrier,and product ion masses), brute-force search methods, such as STAT, may be usedto compare and experimental tandem mass spectrum to those of all possibletheoretical structures. One problem however, is that the number of possiblestructures increases exponentially as the number of monosaccharides in a glycanincreases and thus the search space becomes too big to explore for large glycans.Therefore, this brute-force approach is only feasible where a small number ofglycans are at play. Currently, there isa need for a reconstruction method that has a reduced computational complexity,and a method does not rely on a database of known glycans.
Ourinvention overcomes the drawbacks by providing systems and methods forachieving a de novo method for reconstructing glycan topologies fromexperimental MS data. The de novo method reconstructs possible glycantopologies in a bottom up way by building an interpretation-graph thatinterprets some non-precursor peaks as B or C ions and specifies how tointerpret each B or C ion by appending one or more preceding B and or C ions tomonosaccharide. Additionally, this invention is a machine learning tool thatmay learn fragmentation patterns to assist in selecting the correct glycantopology from a candidate set of proposed structures.
- *IP Issue Date
- None
- *IP Type
- Other Patent
- 国家
- Not Available
- 申请号码
- None
- 国家/地区
- 美国
