Skip to main content

Comparison of transformation, normalization and testing choices in a protein microarray analysis pipeline

Friday, November 08, 2013 — Poster Session IV

2:00 p.m. – 4:00 p.m.

FAES Academic Center (Upper-Level Terrace)




  • J Skinner
  • PD Crompton


Protein microarrays have established an important role in the study of antibody responses to many tropical infectious diseases including brucellosis, malaria, melioidosis and salmonellosis. Papers have been published on appropriate methods for processing and analyzing protein microarray data, but few cover the entire procedure from start to finish or make fair comparisons among the various processing and analytical alternatives. We have categorized the processing of protein microarray data into three distinct steps: transformation, normalization and background subtraction. We also present several statistical testing options to compare the breadth and magnitude of antibody profiles. We then compare several transformation, normalization and statistical testing methods using previously published protein microarray data in which Plasmodium falciparum-specific antibody profiles were examined in a longitudinal cohort study in Kambila, Mali. We find robust linear model (RLM) normalization with generalized linear model tests of the antibody profile breadth to reveal the most powerful insights into the immune response against malaria. Going forward, establishing standard approaches to processing and analyzing protein microarray data will improve the rigor of antibody profiling studies and will facilitate cross-study comparisons.

back to top