Managing a Large-Scale Multiomics Project: A Team Science Case Study in Proteogenomics

Abstract

Highly collaborative scientists are often called on to extend their expertise to different types of projects and to expand the scope and scale of projects well beyond their previous experience. For a large-scale project involving "big data" to be successful, several different aspects of the research plan need to be developed and tested, which include but are not limited to the experimental design, sample collection, sample preparation, metadata recording, technical capability, data acquisition, approaches for data analysis, methods for integration of different data types, recruitment of additional expertise as needed to guide the project, and strategies for clear communication throughout the project. To capture this process, we describe an example project in proteogenomics that built on our collective expertise and experience. Key steps included definition of hypotheses, identification of an appropriate clinical cohort, pilot projects to assess feasibility, refinement of experimental designs, and extensive discussions involving the research team throughout the process. The goal of this chapter is to provide the reader with a set of guidelines to support development of other large-scale multiomics projects.

Keywords: Big data; Biostatistics; Cancer; Experimental design; Informatics; Landscape paper; Planning; Proteogenomics.

Fig. 1

Experimental workflow for lung squamous cell carcinoma proteogenomics. The overall process begins with tissue specimens and proceeds to generate datasets (upper left). To organize a study of this type, multiple identifiers are required to link the samples and data back to the patient (lower left). The overall workflow diagram is presented to show the overall process and emphasize the integrative nature and illustrate the high need for communication and consensus in terms of strategy

Fig. 2

Summary of selected clinical data for lung squamous cell carcinoma patients. Demographics and tumor characterization data (a) are presented with plots of age at diagnosis (b) and overall survival (c)

Fig. 3

Experimental workflow diagram for proteomics sample preparation and data for protein recovery from frozen tumors. The workflow (a) describes the steps in sample processing and digestion. The pooled proteome serves as a control for batch-to-batch comparison of the 29 TMT experiments. Bulk digest is saved for future posttranslational modification (PTM) experiments or quantitative follow-up. Protein assay results are included for each tumor homogenate in chronological order of processing (b)

Fig. 4

Digestion quality control using data independent acquisition LC-MS/MS. The workflow (a) describes data acquisition and associated metadata. The numbers of identified peptides and proteins indicated four samples (arrows) that failed quality control and needed to be digested again (b)

Fig. 5

Tandem mass tag labeling and peptide fractionation for expression proteomics. The workflow (a) describes sample processing, data acquisition, and associated metadata. The amount of cross talk between TMT channels for Batch 16 are shown as an example of the quality control data (b). Overlaid UV chromatograms show the reproducibility of peptide separation across the 29 TMT batches (c)

Fig. 6

Example operational diagram for proteomics weekly lab activities and instrument usage for TMT labeling and quality control. Lab members label peptides with the TMT reagents, acquire LC-MS/MS data for quality control, and perform database searches and quantitative analysis for labeling. One LC-MS/MS instrument is dedicated to perform the experiments during this time