Point and Interval Predictions of Protein Concentrations in ELISA Microarray Assays

• Activity Number: 155
• Type: Contributed
• Date/Time: Monday, August 7, 2006 : 10:30 AM to 12:20 PM
• Sponsor: Biometrics Section
• Author(s): Don S. Daly*+ and Kevin K. Anderson and Amanda M. White and Susan S. Varnum and Richard C. Zangar
• Companies: Pacific Northwest National Laboratory and Pacific Northwest National Laboratory and Pacific Northwest National Laboratory and Pacific Northwest National Laboratory and Pacific Northwest National Laboratory
• Keywords: ELISA microarray immunoassay ; monotonic spline ; propagation of error ; Monte Carlo simulation

Abstract

An ELISA microarray assay predicts simultaneously the concentrations of numerous proteins in a small sample. These predictions are uncertain due to processing error so that prediction error estimates are critical to inferring biological significance and improving the assay. Using real and simulated data, we investigate blends of classic and new methods to make point and interval ELISA concentration predictions: logistic and spline models, NLS and PCLS fits, delta method and Monte Carlo predictors. All blends give acceptable results in most cases. The classic blends are more understandable to biologists but do not always converge and do generate unrealistically divergent intervals. The new blends, though less comprehensible to a layman, are robust and true to the underlying physics. These methods are available as a modular Java/R open source application.