Quantifiable peptide library bridges the gap for proteomics based biomarker discovery and validation on breast cancer.

Mass spectrometry (MS) based proteomics is widely used for biomarker discovery. However, often, most biomarker candidates from discovery are discarded during the validation processes. Such discrepancies between biomarker discovery and validation are caused by several factors, mainly due to the differences in analytical methodology and experimental conditions. Here, we generated a peptide library which allows discovery of biomarkers in the equal settings as the validation process, thereby making the transition from discovery to validation more robust and efficient. The peptide library initiated with a list of 3393 proteins detectable in the blood from public databases. For each protein, surrogate peptides favorable for detection in mass spectrometry was selected and synthesized. A total of 4683 synthesized peptides were spiked into neat serum and plasma samples to check their quantifiability in a 10 min liquid chromatography-MS/MS run time. This led to the PepQuant library, which is composed of 852 quantifiable peptides that cover 452 human blood proteins. Using the PepQuant library, we discovered 30 candidate biomarkers for breast cancer. Among the 30 candidates, nine biomarkers, FN1, VWF, PRG4, MMP9, CLU, PRDX6, PPBP, APOC1, and CHL1 were validated. By combining the quantification values of these markers, we generated a machine learning model predicting breast cancer, showing an average area under the curve of 0.9105 for the receiver operating characteristic curve.

Accurate Prediction of y Ions in Beam-Type Collision-Induced Dissociation Using Deep Learning.

Peptide fragmentation spectra contain critical information for the identification of peptides by mass spectrometry. In this study, we developed an algorithm that more accurately predicts the high-intensity peaks among the peptide spectra. The training data are composed of 180,833 peptides from the National Institute of Standards and Technology and Proteomics Identification database, which were fragmented by either quadrupole time-of-flight or triple-quadrupole collision-induced dissociation methods. Exploratory analysis of the peptide fragmentation pattern was focused on the highest intensity peaks that showed proline, peptide length, and a sliding window of four amino acid combination that can be exploited as key features. The amino acid sequence of each peptide and each of the key features were allocated to different layers of the model, where recurrent neural network, convolutional neural network, and fully connected neural network were used. The trained model, PrAI-frag, accurately predicts the fragmentation spectra compared to previous machine learning-based prediction algorithms. The model excels at high-intensity peak prediction, which is advantageous to selective/multiple reaction monitoring application. PrAI-frag is provided via a Web server which can be used for peptides of length 6-15.