One effective method for examining materials at the nanoscale is small-angle scattering (SAS). However, the fact that it cannot function without some prior knowledge of the chemical composition of a sample has limited its usage in research to this point. A more sophisticated method that combines SAS with machine learning algorithms is presented by Eugen Anitas of the Bogoliubov Laboratory of Theoretical Physics in Dubna, Russia, through recent research that was published in The European Physical Journal E.
Known as α-SAS, this method may examine molecular samples without requiring significant preparation or computational power, and it may help scientists learn more about the characteristics of intricate biomolecules including proteins, lipids, and carbohydrates.
When radiation, usually X-rays or neutrons, interacts with molecular structures suspended in a solvent, SAS monitors the deflection of the radiation. Through manipulation of the solvent’s composition, researchers can increase or decrease the system’s visibility: this is known as “contrast variation.” However, prior conducting the experiment, scientists still need to know certain information about the chemical makeup of the sample in order for this to operate.
Anitas overcome this constraint through his research by combining machine learning algorithms with SAS, resulting in a method known as α-SAS. This method involved doing numerous random simulations of the suspended sample and examining the distribution of the results in order to estimate the outcomes of small-angle neutron scattering (SANS).
Anitas used two distinct case examples to illustrate the potential of α-SAS. Among these, the first one examined “Janus particles”—man-made, self-propelling objects with a well-known intensity of neutron scattering and contrast fluctuation. Second, he used a sophisticated molecular system based on proteins to test the method.
Anitas’ measurements of the molecular structures were far more effective in each instance than they would have been in the absence of any machine learning integration. Anitas is therefore optimistic that, thanks to his method, SAS may soon get even more potent for studying molecular structures in light of these encouraging findings.
