Variance Calculations for Quantitative Real-Time PCR Experiments with Multiple Levels of Replication

Abstract

Heap bioleaching is an established technology for recovering copper from low-grade sulphide ores. Recently, genetics-based approaches have been employed to characterize mineral-processing bacteria. In these approaches, data analysis is a key issue. Consequently, it is of fundamental importance to provide adequate mathematical models and statistical tools to draw reliable conclusions. The present work relates to current studies of the consortium of organisms inhabiting the bioleaching heap of the Escondida mine in Northern Chile. These studies aim to describe and understand the relationship between the dynamics of the community and the performance of the industrial process. Here, we consider a series of quantitative real-time polymerase chain reaction (PCR) experiments performed to quantify six different microorganisms at various stages of the bioleaching cycle. Establishing the reliability of the data obtained by real-time PCR requires the estimation of the error variance at several different levels. The results obtained show that the sampling component of the error variance is the dominant source of variability for most microorganisms. An estimate for the proportional reduction in residual standard deviation from the use of extraction and real-time PCR triplicates was found to range from 3% to 27% for the different organisms. This result suggests that triplicate assays would produce only a modest reduction in error variance compared to more frequent sampling from the heap.