The MicroArray Quality Control (MAQC)-II study of common practices for the development and validation of microarray-based predictive models

Abstract

Gene expression data from microarrays are being applied to predict preclinical and clinical endpoints, but the reliability of these predictions has not been established. In the MAQC-II project, 36 independent teams analyzed six microarray data sets to generate predictive models for classifying a sample with respect to one of 13 endpoints indicative of lung or liver toxicity in rodents, or of breast cancer, multiple myeloma or neuroblastoma in humans. In total, >30,000 models were built using many combinations of analytical methods. The teams generated predictive models without knowing the biological meaning of some of the endpoints and, to mimic clinical reality, tested the models on data that had not been used for training. We found that model performance depended largely on the endpoint and team proficiency and that different approaches generated models of similar performance. The conclusions and recommendations from MAQC-II should be useful for regulatory agencies, study committees and independent investigators that evaluate methods for global gene expression analysis.

A standardized fold change method for microarray differential expression analysis used to reveal genes involved in acute rejection in murine allograft models

“…Microarray is widely used and accepted as a stable, well established and less costly technology to investigate gene expression data 1, 8, 14, 15. In this study based on microarray data, we established a novel method, SFC, to detect differential expression and compared it with the t test and Limma.…”

“…The MAQC project was developed by the US Food and Drug Administration (FDA) to provide standards and quality control metrics and involved six centers [Applied Biosystems (Thermo Fisher Scientific, Waltham, MA, USA), Affymetrix (Santa Clara, CA, USA), Agilent Technologies (Santa Clara, CA, USA), GE Healthcare (Chicago, IL, USA), Illumina (San Diego, CA, USA) and Eppendorf (Hamburg, Germany)] that are major providers of microarray platforms and RNA samples 1, 8. The reproducibility of the top 100 and 1000 significant genes was estimated inter‐ and intra‐platform by the three statistical methods, and heatmaps were drawn with the matrix of each batch.…”

Dual roles for immune metagenes in breast cancer prognosis and therapy prediction

“…The dataset accession numbers are GSE25066, GSE20194, GSE20271, GSE22093 and GSE23988. Specific details of the patient cohorts are described elsewhere [42,[48][49][50][51] and summarized below. All microarray experiments pertaining to these datasets were conducted at the Department of Pathology, MD Anderson Cancer Center (MDACC), Houston, Texas, as part of several international and multicenter studies conducted between 2000 and 2010.…”

“…All microarray experiments pertaining to these datasets were conducted at the Department of Pathology, MD Anderson Cancer Center (MDACC), Houston, Texas, as part of several international and multicenter studies conducted between 2000 and 2010. According to previously published reports [48][49][50][51] for each study the research protocol was approved by one or more institutional review boards, and all participating patients provided written informed consent consistent with the principles of the Declaration of Helsinki. Expression profiles were generated from RNA samples isolated from fine needle aspirates (FNAs) or needle core biopsies of breast tumors (stage I to III) collected prior to treatment with neoadjuvant chemotherapy.…”

Statistical modeling for differential transcriptome analysis using RNA-Seq technology

“…Additionally, databases like SEQC (SEquencing Quality Control) have been established to assess the performance of the NGS technologies. SEQC, also known as MAQC-III (the third phase of the MAQC project), is a follow up from the MAQC and MAQC-II projects [47,48] . It aims at assessing the technical reproducibility of NGS technologies such as RNA-Seq by generating benchmark datasets with known reference samples.…”