Limits...
MicroRNA array normalization: an evaluation using a randomized dataset as the benchmark.

Qin LX, Zhou Q - PLoS ONE (2014)

Bottom Line: The non-randomized dataset was assessed for differential expression after normalization and compared against the benchmark.Normalization improved the true positive rate significantly in the non-randomized data but still possessed a false discovery rate as high as 50%.We concluded the paper with some insights on possible causes of false discoveries to shed light on how to improve normalization for microRNA arrays.

View Article: PubMed Central - PubMed

Affiliation: Department of Epidemiology and Biostatistics, Memorial Sloan Kettering Cancer Center, New York, New York, United States of America.

ABSTRACT
MicroRNA arrays possess a number of unique data features that challenge the assumption key to many normalization methods. We assessed the performance of existing normalization methods using two microRNA array datasets derived from the same set of tumor samples: one dataset was generated using a blocked randomization design when assigning arrays to samples and hence was free of confounding array effects; the second dataset was generated without blocking or randomization and exhibited array effects. The randomized dataset was assessed for differential expression between two tumor groups and treated as the benchmark. The non-randomized dataset was assessed for differential expression after normalization and compared against the benchmark. Normalization improved the true positive rate significantly in the non-randomized data but still possessed a false discovery rate as high as 50%. Adding a batch adjustment step before normalization further reduced the number of false positive markers while maintaining a similar number of true positive markers, which resulted in a false discovery rate of 32% to 48%, depending on the specific normalization method. We concluded the paper with some insights on possible causes of false discoveries to shed light on how to improve normalization for microRNA arrays.

Show MeSH

Related in: MedlinePlus

Scatter plot comparing pooled standard deviations in the benchmark data and that in the test data for (A) no normalization, (B) median normalization, and (C) quantile normalization.Black “x”: true positive markers. Red “x”: false positive markers. Blue “x”: false negative markers. Black dots: true negative markers.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4048305&req=5

pone-0098879-g003: Scatter plot comparing pooled standard deviations in the benchmark data and that in the test data for (A) no normalization, (B) median normalization, and (C) quantile normalization.Black “x”: true positive markers. Red “x”: false positive markers. Blue “x”: false negative markers. Black dots: true negative markers.

Mentions: Figure 2A and Figure 3A show that array effects led to both a (dominantly negative) bias in mean differences (that is, ovarian mean minus endometrial mean) and an overall increase in variability in the test data. More specifically, the bias primarily shifted the data towards endometrial tumors: it pulled markers whose true mean differences are around zero away from zero, and some markers whose true mean differences are moderately positive close to zero. Most false positive markers had mean differences close to zero in the benchmark and were resulted from the negative biases in mean difference caused by array effects. Most false negative markers had positive mean differences in the benchmark and were resulted from the under-estimated magnitudes of mean difference. The level of increase in data variability is similar for the majority of markers. This increase partly contributed to the occurrence of false negative markers.


MicroRNA array normalization: an evaluation using a randomized dataset as the benchmark.

Qin LX, Zhou Q - PLoS ONE (2014)

Scatter plot comparing pooled standard deviations in the benchmark data and that in the test data for (A) no normalization, (B) median normalization, and (C) quantile normalization.Black “x”: true positive markers. Red “x”: false positive markers. Blue “x”: false negative markers. Black dots: true negative markers.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4048305&req=5

pone-0098879-g003: Scatter plot comparing pooled standard deviations in the benchmark data and that in the test data for (A) no normalization, (B) median normalization, and (C) quantile normalization.Black “x”: true positive markers. Red “x”: false positive markers. Blue “x”: false negative markers. Black dots: true negative markers.
Mentions: Figure 2A and Figure 3A show that array effects led to both a (dominantly negative) bias in mean differences (that is, ovarian mean minus endometrial mean) and an overall increase in variability in the test data. More specifically, the bias primarily shifted the data towards endometrial tumors: it pulled markers whose true mean differences are around zero away from zero, and some markers whose true mean differences are moderately positive close to zero. Most false positive markers had mean differences close to zero in the benchmark and were resulted from the negative biases in mean difference caused by array effects. Most false negative markers had positive mean differences in the benchmark and were resulted from the under-estimated magnitudes of mean difference. The level of increase in data variability is similar for the majority of markers. This increase partly contributed to the occurrence of false negative markers.

Bottom Line: The non-randomized dataset was assessed for differential expression after normalization and compared against the benchmark.Normalization improved the true positive rate significantly in the non-randomized data but still possessed a false discovery rate as high as 50%.We concluded the paper with some insights on possible causes of false discoveries to shed light on how to improve normalization for microRNA arrays.

View Article: PubMed Central - PubMed

Affiliation: Department of Epidemiology and Biostatistics, Memorial Sloan Kettering Cancer Center, New York, New York, United States of America.

ABSTRACT
MicroRNA arrays possess a number of unique data features that challenge the assumption key to many normalization methods. We assessed the performance of existing normalization methods using two microRNA array datasets derived from the same set of tumor samples: one dataset was generated using a blocked randomization design when assigning arrays to samples and hence was free of confounding array effects; the second dataset was generated without blocking or randomization and exhibited array effects. The randomized dataset was assessed for differential expression between two tumor groups and treated as the benchmark. The non-randomized dataset was assessed for differential expression after normalization and compared against the benchmark. Normalization improved the true positive rate significantly in the non-randomized data but still possessed a false discovery rate as high as 50%. Adding a batch adjustment step before normalization further reduced the number of false positive markers while maintaining a similar number of true positive markers, which resulted in a false discovery rate of 32% to 48%, depending on the specific normalization method. We concluded the paper with some insights on possible causes of false discoveries to shed light on how to improve normalization for microRNA arrays.

Show MeSH
Related in: MedlinePlus