Limits...
Evaluation of time profile reconstruction from complex two-color microarray designs.

Fierro AC, Thuret R, Engelen K, Bernot G, Marchal K, Pollet N - BMC Bioinformatics (2008)

Bottom Line: Methods with a dye effect seemed more robust against array failure.Including a dye effect such as in the methods lmbr_dye, anovaFix and anovaMix compensates for residual dye related inconsistencies in the data and renders the results more robust against array failure.Including random effects requires more parameters to be estimated and is only advised when a design is used with a sufficient number of replicates.

View Article: PubMed Central - HTML - PubMed

Affiliation: CNRS UMR 8080, Laboratoire Développement et Evolution, Bat 445, F-91405 Orsay, France. carolina.fierro@gmail.com

ABSTRACT

Background: As an alternative to the frequently used "reference design" for two-channel microarrays, other designs have been proposed. These designs have been shown to be more profitable from a theoretical point of view (more replicates of the conditions of interest for the same number of arrays). However, the interpretation of the measurements is less straightforward and a reconstruction method is needed to convert the observed ratios into the genuine profile of interest (e.g. a time profile). The potential advantages of using these alternative designs thus largely depend on the success of the profile reconstruction. Therefore, we compared to what extent different linear models agree with each other in reconstructing expression ratios and corresponding time profiles from a complex design.

Results: On average the correlation between the estimated ratios was high, and all methods agreed with each other in predicting the same profile, especially for genes of which the expression profile showed a large variance across the different time points. Assessing the similarity in profile shape, it appears that, the more similar the underlying principles of the methods (model and input data), the more similar their results. Methods with a dye effect seemed more robust against array failure. The influence of a different normalization was not drastic and independent of the method used.

Conclusion: Including a dye effect such as in the methods lmbr_dye, anovaFix and anovaMix compensates for residual dye related inconsistencies in the data and renders the results more robust against array failure. Including random effects requires more parameters to be estimated and is only advised when a design is used with a sufficient number of replicates. Because of this, we believe lmbr_dye, anovaFix and anovaMix are most appropriate for practical use.

Show MeSH
Spike-in expression ratio estimates. Reconstructed expression ratio estimates of spikes 7 (+ markers) and 8 (x markers) are plotted for lmbr/lmbr_dye/limmaQual (solid line), anovaFix (dotted line), and anovaMix (dashed line). Concentrations (cpc; copies per cell) of 104 cpc (panel A), 10 cpc (panel B), and 10-1 cpc (panel C) were used as reference points. Estimated ratios were sorted from low to high concentrations. The solid grey line (o-markers) corresponds to the expected ratios for the known concentrations.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC2265676&req=5

Figure 3: Spike-in expression ratio estimates. Reconstructed expression ratio estimates of spikes 7 (+ markers) and 8 (x markers) are plotted for lmbr/lmbr_dye/limmaQual (solid line), anovaFix (dotted line), and anovaMix (dashed line). Concentrations (cpc; copies per cell) of 104 cpc (panel A), 10 cpc (panel B), and 10-1 cpc (panel C) were used as reference points. Estimated ratios were sorted from low to high concentrations. The solid grey line (o-markers) corresponds to the expected ratios for the known concentrations.

Mentions: As lmbr and lmbr_dye and limmaQual gave exactly the same results using this balanced design, we further assessed to what extent lmbr, anovaFix and anovaMix agreed with each other. Fig. 3 shows the effect of using different spike concentrations as reference points for ratio estimation. Panels A through C reflect decreasing reference concentrations. The choice of reference has little effect on the shape of the profile (as indicated by consistent relationships between the different estimates). However, Fig. 3 illustrates that 1) lower reference concentrations (intensities) introduce a bias in the profile (true ratio's are consistently underestimated), 2) irrespective of the concentration of the reference ratio's derived for the lower expression values of the test are nearly identical, and thus uninformative. Both observations can be attributed to the lower saturation characteristics of microarray data (low concentrations do not generate signals that are distinguishable from the background). Although not as complex as the previously used loop or interwoven designs, the spiked-in design illustrates that this lower saturation effect, an inherent property of microarray data, can distort estimated profiles: interpretation of ratios with lower signals for test or reference should be done with care.


Evaluation of time profile reconstruction from complex two-color microarray designs.

Fierro AC, Thuret R, Engelen K, Bernot G, Marchal K, Pollet N - BMC Bioinformatics (2008)

Spike-in expression ratio estimates. Reconstructed expression ratio estimates of spikes 7 (+ markers) and 8 (x markers) are plotted for lmbr/lmbr_dye/limmaQual (solid line), anovaFix (dotted line), and anovaMix (dashed line). Concentrations (cpc; copies per cell) of 104 cpc (panel A), 10 cpc (panel B), and 10-1 cpc (panel C) were used as reference points. Estimated ratios were sorted from low to high concentrations. The solid grey line (o-markers) corresponds to the expected ratios for the known concentrations.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC2265676&req=5

Figure 3: Spike-in expression ratio estimates. Reconstructed expression ratio estimates of spikes 7 (+ markers) and 8 (x markers) are plotted for lmbr/lmbr_dye/limmaQual (solid line), anovaFix (dotted line), and anovaMix (dashed line). Concentrations (cpc; copies per cell) of 104 cpc (panel A), 10 cpc (panel B), and 10-1 cpc (panel C) were used as reference points. Estimated ratios were sorted from low to high concentrations. The solid grey line (o-markers) corresponds to the expected ratios for the known concentrations.
Mentions: As lmbr and lmbr_dye and limmaQual gave exactly the same results using this balanced design, we further assessed to what extent lmbr, anovaFix and anovaMix agreed with each other. Fig. 3 shows the effect of using different spike concentrations as reference points for ratio estimation. Panels A through C reflect decreasing reference concentrations. The choice of reference has little effect on the shape of the profile (as indicated by consistent relationships between the different estimates). However, Fig. 3 illustrates that 1) lower reference concentrations (intensities) introduce a bias in the profile (true ratio's are consistently underestimated), 2) irrespective of the concentration of the reference ratio's derived for the lower expression values of the test are nearly identical, and thus uninformative. Both observations can be attributed to the lower saturation characteristics of microarray data (low concentrations do not generate signals that are distinguishable from the background). Although not as complex as the previously used loop or interwoven designs, the spiked-in design illustrates that this lower saturation effect, an inherent property of microarray data, can distort estimated profiles: interpretation of ratios with lower signals for test or reference should be done with care.

Bottom Line: Methods with a dye effect seemed more robust against array failure.Including a dye effect such as in the methods lmbr_dye, anovaFix and anovaMix compensates for residual dye related inconsistencies in the data and renders the results more robust against array failure.Including random effects requires more parameters to be estimated and is only advised when a design is used with a sufficient number of replicates.

View Article: PubMed Central - HTML - PubMed

Affiliation: CNRS UMR 8080, Laboratoire Développement et Evolution, Bat 445, F-91405 Orsay, France. carolina.fierro@gmail.com

ABSTRACT

Background: As an alternative to the frequently used "reference design" for two-channel microarrays, other designs have been proposed. These designs have been shown to be more profitable from a theoretical point of view (more replicates of the conditions of interest for the same number of arrays). However, the interpretation of the measurements is less straightforward and a reconstruction method is needed to convert the observed ratios into the genuine profile of interest (e.g. a time profile). The potential advantages of using these alternative designs thus largely depend on the success of the profile reconstruction. Therefore, we compared to what extent different linear models agree with each other in reconstructing expression ratios and corresponding time profiles from a complex design.

Results: On average the correlation between the estimated ratios was high, and all methods agreed with each other in predicting the same profile, especially for genes of which the expression profile showed a large variance across the different time points. Assessing the similarity in profile shape, it appears that, the more similar the underlying principles of the methods (model and input data), the more similar their results. Methods with a dye effect seemed more robust against array failure. The influence of a different normalization was not drastic and independent of the method used.

Conclusion: Including a dye effect such as in the methods lmbr_dye, anovaFix and anovaMix compensates for residual dye related inconsistencies in the data and renders the results more robust against array failure. Including random effects requires more parameters to be estimated and is only advised when a design is used with a sufficient number of replicates. Because of this, we believe lmbr_dye, anovaFix and anovaMix are most appropriate for practical use.

Show MeSH