Development of Peptide\nIdentification System for ToF-SIMS\nSpectra Using Supervised Machine Learning

Time-of-flight secondary ion mass spectrometry (ToF-SIMS)\ndata\ninterpretation for organic materials is complicated because of various\nfragment ions produced from each molecule and the overlapping of certain\nmass peaks from different molecules. Fragmentation mechanisms in SIMS\nare complex because different sputtering and ionization processes\ncan simultaneously occur. Therefore, a prediction system that can\nidentify materials in a sample is required. A novel prediction system\nfor peptides based on ToF-SIMS and amino-acid-based teaching information\n(labels) for supervised machine learning was developed. To develop\nthe prediction system for general organic materials, the annotation\nof materials is crucial to creating effective labels for supervised\nlearning. Peptides are composed of 20 amino acid residues, which\ncan be used as labels. We previously developed a peptide prediction\nsystem using Random Forest, a supervised machine-learning method.\nHowever, only the amino acids contained in the target peptide were\npredicted, and the amino acid sequence was unable to be assumed. In\nthis study, the amino acid sequence of the test peptide was determined\nby adding the information on two adjacent amino acids to the labels.\nOnce the prediction system learned the target peptide spectra, the\npeptides in the newly obtained ToF-SIMS spectra could be identified.\nThe new prediction system also provides useful information for the\nidentification of unknown peptides. The prediction results indicate\nthat two adjacent permutations of amino acids are effective pieces\nof teaching information for expressing the amino acid sequence of\na peptide.