Expert Review Introduction

Targeted MS experiments have traditionally been of relatively smaller scale than discovery experiments,yet the data analysis can be prone to high rates of incorrect integration boundary determination. With the introduction of new MS hardware and acquisition strategies, notably Stellar MS with Adaptive RT and PRM Conductor, targeted experiments can routinely exceed 1000 targets. This increased scale necessitates improved tools for data analysis. Expert review is a software tool that improves the LC peak boundary imputation process, through the use of replicate-level and experiment-level analysis. The methodologies used were inspired by our algorithm used for real-time retention time alignment, known as Adaptive RT.

• To use Expert Review, you should have a reference set of data, where the all the peaks have been picked correctly. If you are doing a label-free experiment, the easiest way to obtain a reference file is to have a set of discovery data with compound IDs in it, ensuring that the correct peak was initially selected in Skyline. If the experiment you are doing uses heavy standards, then the reference data can come from injections of the neat standards.
• Expert Review utilizes two levels of analysis, a Replicate-level scoring based on an analysis of each replicate in isolation, and an Experiment-level re-scoring for each precursor, that considers all of its replicates.

Expert review has mostly been used for analysis of dilution curves, which are challenging test cases for integration boundary determination, because by design there are many replicates where the analytes of interest are diluted to below limits of detection. We have found via manual inspection in many cases that Expert Review makes no errors, and those errors that do happen are typically either very small, or caused by the use of the RT outlier correction, which is not suitable for abrupt RT changes, and is an option that can be turned off.