SPECIALLY COLLECTED PAPERS
Shimadzu Review 78[3・4] (2021)
Abstract
The peak picking process is used to determine peak regions in chromatograms. In biotechnology, peak picking is used as basic technology for obtaining information about the quantities of chemical compounds contained in analytical samples. In recent years, many measures for improving peak picking performance have been reported, but none have resulted in faster picking speeds. Therefore, we developed Peakintelligence peak picking software that uses deep learning technology to allow anyone to easily obtain results equivalent to an expert. First, chromatograms acquired from food and biological samples were used to augment the training data. Then parameter-free peak picking software was developed by building a peak picking network based on that augmented training data and validation data. Next, using test data that was separate from the training data, the modification time was evaluated. That confirmed that Peakintelligence requires a shorter modification time than two conventional software programs. The results show that Peakintelligence shortens the time required for peak picking, which can lead to resolving that bottleneck in chromatography-based quantitative analysis work.
1IT Solutions Unit, Analytical & Measuring Instruments Division, Shimadzu Corporation, Kyoto, Japan
2AI Solution Unit, Technology Research Laboratory, Shimadzu Corporation, Kyoto, Japan
3MS Business Unit, Life Science Business Department, Analytical & Measuring Instruments Division, Shimadzu Corporation, Kyoto, Japan
4Life Science Business Department, Analytical & Measuring Instruments Division, Shimadzu Corporation, Kyoto, Japan
5AI Services Department, DX Services Division, Fujitsu Limited, Tokyo, Japan
6Graduate School of Engineering, Osaka University, Osaka, Japan
7Osaka University Shimadzu Omics Innovation Research Laboratories, Osaka University, Osaka, Japan
*The information contained in Shimadzu Review has not been modified since the original publication date. Please be aware that in some cases, products mentioned within the articles are no longer available.