Machine learning applications in proteomics research: How the past can boost the future

Abstract

Machine learning is a subdiscipline within artificial intelligence that focuses on algorithms that allow computers to learn solving a (complex) problem from existing data. This ability can be used to generate a solution to a particularly intractable problem, given that enough data are available to train and subsequently evaluate an algorithm on. Since MS‐based proteomics has no shortage of complex problems, and since publicly available data are becoming available in ever growing amounts, machine learning is fast becoming a very popular tool in the field. We here therefore present an overview of the different applications of machine learning in proteomics that together cover nearly the entire wet‐ and dry‐lab workflow, and that address key bottlenecks in experiment planning and design, as well as in data processing and analysis.

Advances in proteomics for production strain analysis

“…This has led to the use of cutting-edge informatics, such as machine learning, being employed in analysis of the data [41].…”

Designing biomedical proteomics experiments: state-of-the-art and future perspectives

“…Although the prediction of peptide LC retention times might struggle with the large variety across LC methods and analyses, predictive models have been built for chromatography in proteomics, helping experimentalists by providing expected elution times or hydrophobicity indices (in case of reverse phase chromatography) [86]. While the first RT prediction models assumed that peptide RT is a linear function of the amino acid sequence [90], more recent models also focus on peptide length or positional effects of the amino acid residues [91,92].…”

“…This information is crucial as only these peptides will be analyzed via LC-MS and thus can be detected in the experiment. Predictive proteolysis models can thus improve identification rates in shotgun proteomics and/or provide a priori prediction of suitable peptides for targeted proteomics analyses [86]. Software predicting cleavage probabilities exists for many proteases [86], with as usage mode the theoretical digestion of a single protein or mixture.…”

“…Predictive proteolysis models can thus improve identification rates in shotgun proteomics and/or provide a priori prediction of suitable peptides for targeted proteomics analyses [86]. Software predicting cleavage probabilities exists for many proteases [86], with as usage mode the theoretical digestion of a single protein or mixture. For trypsin, for example, Cleaving prediction with decision trees (CP-DT) uses the positional information of amino acid sequences around the tryptic site to estimate whether or not the protein will be cleaved [88].…”

“…Large-scale SRM assays, where hundreds of peptides need to be targeted, require even MS instrument parameters that work well with the broad diversity of peptides to be targeted. Software packages and computational studies attempt to improve or assist determining these parameters for a given experiment [86]. These efforts start with modeling the proteomics pipeline [87] to optimize some experimental parameters.…”