Section: New Results

Modeling Macro-molecular Assemblies

Macro-molecular assembly, reconstruction by data integration, proteomics, modeling with uncertainties, curved Voronoi diagrams, topological persistence.

Stoichiometry Determination for Mass-spectrometry Data: the Interval Case

Participants : Deepesh Agarwal, Frédéric Cazals, Noël Malod-Dognin.

In structural proteomics, given the individual masses of a set of protein types and the exact mass of a protein complex, the exact stoichiometry determination problem (SD), also known as the money-change problem, consists of enumerating all the stoichiometries of these types which allow to recover the target mass. If the target mass suffers from experimental uncertainties, the interval SD problem consists of finding all the stoichiometry vectors compatible with a target mass within an interval.

We make contributions in two directions [18] . From a theoretical standpoint, we present a constant-memory space algorithm (DIOPHANTINE ) and an output sensitive dynamic programming based algorithm (DP++ ), both inherently addressing the interval SD problem. From an applied perspective, we raise three points. First, we show that DIOPHANTINE and DP++ yield an improvement from 3 to 4 orders of magnitude over state-of-the-art exact SD algorithms, for typical protein complexes facing uncertainties on the target mass in the range 0.1-1%. Second, we show that DIOPHANTINE behaves like an output-sensitive algorithm—especially when the interval width increases, albeit such a property cannot be expected in general. Third, from a biological perspective, using a panel of biological complexes (eukaryotic translation factor, yeast exosome, 19S proteasome sub-unit, nuclear pore complex), we stress the importance of enumeration, even at a null noise level.

The programs accompanying this paper are available from http://team.inria.fr/abs/addict/ .