TY - GEN
T1 - Assessing agreement between microRNA microarray platforms via linear measurement error models
AU - Bassani, Niccolò
AU - Ambrogi, Federico
AU - Biganzoli, Elia
PY - 2013
Y1 - 2013
N2 - Over the last years miRNA microarray platforms have provided insights in the biological mechanisms underlying onset and development of several diseases and have thus become a very popular instrument for profiling thousands of miRNA simultaneously. However, because of large variety of microarray platforms available, an assessment of their performance in terms of both within-platform reliability and between-platform agreement is needful. In particular, assessment of platform concordance has been a very relevant issue in the past decade. To date, only a few studies have evaluated this problem in the field of miRNA microarray, and mostly by using improper statistical methods such as the Pearson and Spearman correlation coefficients. In this work we suggest to use a recently proposed modified version of the classical Bland-Altman approach for comparing clinical measurement methods. This modified version is useful in that allows not only to evaluate agreement between different miRNA microarray platforms, but also to assess which are the potential sources of disagreement/bias between them. Two samples were profiled using Affymetrix, Agilent and Illumina miRNA platform using three technical replicates each, and pairwise agreement between platforms was evaluated within each sample. Our results suggest that, after bias correction, Illumina and Agilent show the best patterns of agreement for both samples involved in the experiment, whereas Affymetrix is the one which seem to "disagree" most, suggesting that a linear relationship as that hypothesized by the measurement error model used is not able to capture the complexity of the phenomenon. In the future it will be interesting to apply this method also to the comparison of microarray and NGS platform, a topic which is becoming more and more relevant, also by adopting non-linear measurement error models to depict relationships between platforms.
AB - Over the last years miRNA microarray platforms have provided insights in the biological mechanisms underlying onset and development of several diseases and have thus become a very popular instrument for profiling thousands of miRNA simultaneously. However, because of large variety of microarray platforms available, an assessment of their performance in terms of both within-platform reliability and between-platform agreement is needful. In particular, assessment of platform concordance has been a very relevant issue in the past decade. To date, only a few studies have evaluated this problem in the field of miRNA microarray, and mostly by using improper statistical methods such as the Pearson and Spearman correlation coefficients. In this work we suggest to use a recently proposed modified version of the classical Bland-Altman approach for comparing clinical measurement methods. This modified version is useful in that allows not only to evaluate agreement between different miRNA microarray platforms, but also to assess which are the potential sources of disagreement/bias between them. Two samples were profiled using Affymetrix, Agilent and Illumina miRNA platform using three technical replicates each, and pairwise agreement between platforms was evaluated within each sample. Our results suggest that, after bias correction, Illumina and Agilent show the best patterns of agreement for both samples involved in the experiment, whereas Affymetrix is the one which seem to "disagree" most, suggesting that a linear relationship as that hypothesized by the measurement error model used is not able to capture the complexity of the phenomenon. In the future it will be interesting to apply this method also to the comparison of microarray and NGS platform, a topic which is becoming more and more relevant, also by adopting non-linear measurement error models to depict relationships between platforms.
KW - agreement
KW - Bland-Altman
KW - measurement error model
KW - microarrays
KW - microRNA
UR - http://www.scopus.com/inward/record.url?scp=84883322738&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84883322738&partnerID=8YFLogxK
U2 - 10.1007/978-3-642-38342-7_11
DO - 10.1007/978-3-642-38342-7_11
M3 - Conference contribution
AN - SCOPUS:84883322738
SN - 9783642383410
VL - 7845 LNBI
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 117
EP - 131
BT - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
T2 - 9th International Meeting on Computational Intelligence Methods for Bioinformatics and Biostatistics, CIBB 2012
Y2 - 12 July 2012 through 14 July 2012
ER -