Article
Biochemical Research Methods
Daniel P. Zolg, Siegfried Gessulat, Carmen Paschke, Michael Graber, Magnus Rathke-Kuhnert, Florian Seefried, Kai Fitzemeier, Frank Berg, Daniel Lopez-Ferrer, David Horn, Christoph Henrich, Andreas Huhmer, Bernard Delanghe, Martin Frejno
Summary: INFERYS rescoring utilizes deep learning for accurate prediction of peptide fragment ion intensities, enhancing the scoring accuracy of Sequest HT search engine results. By combining intensity-based scores with classical search engine scores, this approach leads to improved peptide spectrum match and identification numbers in proteomic data analysis.
RAPID COMMUNICATIONS IN MASS SPECTROMETRY
(2021)
Article
Multidisciplinary Sciences
Weiping Sun, Qianqiu Zhang, Xiyue Zhang, Ngoc Hieu Tran, M. Ziaur Rahman, Zheng Chen, Chao Peng, Jun Ma, Ming Li, Lei Xin, Baozhen Shan
Summary: GlycanFinder is a database search and de novo sequencing tool for accurate identification of intact glycopeptides. It integrates peptide-based and glycan-based search strategies to overcome the challenge of complex fragmentation. With a deep learning model, GlycanFinder is able to sequence glycans not found in the database. Extensive analyses and benchmarking studies demonstrate its comparable performance to leading software and its capability to identify previously unknown glycopeptides.
NATURE COMMUNICATIONS
(2023)
Article
Chemistry, Analytical
Taran Driver, Nikhil Bachhawat, Ruediger Pipkorn, Leszek J. Frasinski, Jon P. Marangos, Marina Edelson-Averbukh, Vitali Averbukh
Summary: The study introduces a protein database search engine based on 2D-PC-MS method, which can accurately identify protein sequences by matching theoretical and experimentally detected correlating fragments, with high structural specificity. This search engine is not only suitable for peptide identification, but also for intact protein identification.
ANALYTICAL CHEMISTRY
(2021)
Article
Computer Science, Artificial Intelligence
Shaohua Zhao, Hua Zhang, Xin Zhang, Wenmin Li, Fei Gao, Qiaoyan Wen
Summary: This paper proposes a forward privacy multikeyword ranked search scheme over encrypted databases, which protects user privacy and achieves search result ranking by adding dummy elements.
INTERNATIONAL JOURNAL OF INTELLIGENT SYSTEMS
(2022)
Article
Biochemistry & Molecular Biology
Kevin McDonnell, Enda Howley, Florence Abram
Summary: This research compared the performance of two state-of-the-art de novo peptide sequencing algorithms, Novor and DeepNovo, with a focus on their handling of missing fragmentation cleavage sites and noise. The study found that DeepNovo performed better overall than Novor, but Novor recalled more correct amino acids when 6 or more cleavage sites were missing.
COMPUTATIONAL AND STRUCTURAL BIOTECHNOLOGY JOURNAL
(2022)
Article
Biochemistry & Molecular Biology
Kevin McDonnell, Enda Howley, Florence Abram
Summary: Proteomics is a technique used to study system-wide protein expression, which has wide ranging applications and impacts every area of biology. De novo peptide sequencing, a popular method, is improving with the integration of machine learning. This research evaluates two algorithms for de novo peptide sequencing and explores the characteristics of tandem mass spectra. The study highlights the challenges of missing cleavage sites and noise, and provides recommendations for algorithm improvements.
COMPUTATIONAL AND STRUCTURAL BIOTECHNOLOGY JOURNAL
(2022)
Article
Mathematics
Ibrahim Al-Shourbaji, Na Helian, Yi Sun, Samah Alshathri, Mohamed Abd Elaziz
Summary: This paper discusses the importance of feature selection in the telecommunications industry for machine learning models. It introduces a new approach that combines ant colony optimization and reptile search algorithm, and evaluates its performance in customer churn prediction.
Article
Biochemical Research Methods
Chen Zhou, Shuaijian Dai, Yuanqiao Lin, Sheng Lian, Xiaodan Fan, Ning Li, Weichuan Yu
Summary: In this paper, a comprehensive cross-linking search method called ECL-PF is proposed for cleavable XL-MS data analysis, which can improve the sensitivity of protein-protein interaction detection and protein structure probing.
JOURNAL OF PROTEOME RESEARCH
(2023)
Article
Green & Sustainable Science & Technology
Chayoung Kim, Taejung Park
Summary: This study aims to identify the key factors that influence the actual learning intention leading to participation in adult education. Using longitudinal big data from Korean adults (2017-2020), a predictive model was developed using tree-based machine learning. The results revealed that self-pay education expenses and the highest level of education completed were the most influential variables in predicting the likelihood of lifelong education participation. After grid search, the importance of these variables as well as overall figures, including the false positive rate, improved. Future studies could further enhance the performance of the machine learning model by adjusting hyperparameters using less computational methods.
Article
Mathematics
Mohamed Abd Elaziz, Laith Abualigah, Dalia Yousri, Diego Oliva, Mohammed A. A. Al-Qaness, Mohammad H. Nadimi-Shahraki, Ahmed A. Ewees, Songfeng Lu, Rehab Ali Ibrahim
Summary: Feature selection is a crucial step in soft computing and machine learning algorithms that aims to determine relevant features and improve data processing efficiency. Methods based on metaheuristic techniques have shown better performance compared to traditional methods, with the Atomic Orbital Search as a new approach in this domain.
Article
Economics
Yufei Xia, Lingyun He, Yinguo Li, Yating Fu, Yixin Xu
Summary: A novel dynamic credit scoring model, SurvXGBoost, is proposed by combining survival analysis and GBDT approach, which is expected to enhance predictability compared to statistical survival models. Empirical results demonstrate that SurvXGBoost outperforms other benchmark models on a real-world dataset, particularly in terms of misclassification cost.
TECHNOLOGICAL AND ECONOMIC DEVELOPMENT OF ECONOMY
(2021)
Article
Biochemical Research Methods
Xiangyuan Zeng, Bin Ma
Summary: MSTracer is a new software tool for detecting peptide features from MS data, which incorporates two scoring functions based on machine learning and has demonstrated significantly better performance compared to existing tools.
JOURNAL OF PROTEOME RESEARCH
(2021)
Article
Multidisciplinary Sciences
Margaret Amerley Amarh, Michael Konney Laryea, Lawrence Sheringham Borquaye
Summary: The phenomenon of antimicrobial resistance poses a threat to our ability to combat common infections. This study aimed to design low-toxicity antimicrobial peptides using database filtering technology and evaluate their bioactivities. The designed peptides demonstrated antimicrobial activity and synergistic effects when combined with certain antibiotics.
Article
Biochemical Research Methods
Seungjin Na, Hyunjin Choi, Eunok Paek
Summary: In this study, a predicted Spectral DataBase (pSDB) search strategy called Deephos was proposed to improve the identification of TMT-labeled phosphopeptides in MS/MS spectra. By utilizing deep learning-based fragment ion prediction, a pSDB of TMT-labeled phosphopeptides was compiled. The study also discussed the generation of decoy spectra for more accurate estimation of false discovery rate (FDR). The utilities of Deephos were demonstrated in the analysis of three cancer phosphoproteomes.
Article
Biochemical Research Methods
Tobias Greisager Rehfeldt, Konrad Krawczyk, Mathias Bogebjerg, Veit Schwammle, Richard Rottger
Summary: The MS2AI pipeline automates the process of gathering large quantities of MS data for machine learning applications, addressing three major limitations in ML within the LC-MS field.