Interpretation of the DOME Recommendations for Machine Learning in Proteomics and Metabolomics

Article Biochemical Research Methods

MS2AI: automated repurposing of public peptide LC-MS data for machine learning applications

Tobias Greisager Rehfeldt et al.

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.

BIOINFORMATICS (2022)

Add to Collection

Article Mathematical & Computational Biology

The Matthews correlation coefficient (MCC) is more reliable than balanced accuracy, bookmaker informedness, and markedness in two-class confusion matrix evaluation

Davide Chicco et al.

Summary: In this manuscript, the authors reaffirm the robustness of the Matthews correlation coefficient (MCC) as a metric for summarizing classifier performance. They compare MCC to other metrics such as balanced accuracy, bookmaker informedness, and markedness, and demonstrate situations where MCC provides a more informative response. The authors also identify three exceptions where bookmaker informedness may be more appropriate, but overall suggest MCC as a standard measure for evaluating binary classifications in various fields.

BIODATA MINING (2021)

Add to Collection

Editorial Material Biochemical Research Methods

The AIMe registry for artificial intelligence in biomedical research

Julian Matschinske et al.

Summary: The AIMe registry is a community-driven reporting platform for AI in biomedicine, aiming to improve the accessibility, reproducibility, and usability of biomedical AI models, and allowing future revisions by the community.

NATURE METHODS (2021)

Add to Collection

Editorial Material Biochemical Research Methods

DOME: recommendations for supervised machine learning validation in biology

Ian Walsh et al.

Summary: DOME is a set of community-wide recommendations for reporting supervised machine learning-based analyses applied to biological studies, with the goal of improving machine learning assessment and reproducibility.

NATURE METHODS (2021)

Add to Collection

Article Biochemical Research Methods

Universal Spectrum Identifier for mass spectra

Eric W. Deutsch et al.

Summary: The Universal Spectrum Identifier (USI) provides a standardized mechanism for encoding virtual paths to mass spectra in public repositories, enabling greater transparency and traceability of spectral evidence. Over 1 billion USI identifications from more than 3 billion spectra are already available through ProteomeXchange repositories, supporting the findings of mass spectrometry proteomics studies.

NATURE METHODS (2021)

Add to Collection

Article Computer Science, Information Systems