Limits...
A Method and On-Line Tool for Maximum Likelihood Calibration of Immunoblots and Other Measurements That Are Quantified in Batches.

Andrews SS, Rutherford S - PLoS ONE (2016)

Bottom Line: The conventional approach has two steps: the experimenter deduces a conversion function using measurements on standards and then calibrates (or normalizes) measurements on unknown samples with this function.It returns the most probable values for the sample compositions under the assumptions of a statistical model, making them the maximum likelihood predictors.It is less sensitive to measurement error on standards and enables use of some batches that do not include standards.

View Article: PubMed Central - PubMed

Affiliation: Basic Sciences Division, Fred Hutchinson Cancer Research Center, Seattle, WA, United States of America.

ABSTRACT
Experimental measurements require calibration to transform measured signals into physically meaningful values. The conventional approach has two steps: the experimenter deduces a conversion function using measurements on standards and then calibrates (or normalizes) measurements on unknown samples with this function. The deduction of the conversion function from only the standard measurements causes the results to be quite sensitive to experimental noise. It also implies that any data collected without reliable standards must be discarded. Here we show that a "1-step calibration method" reduces these problems for the common situation in which samples are measured in batches, where a batch could be an immunoblot (Western blot), an enzyme-linked immunosorbent assay (ELISA), a sequence of spectra, or a microarray, provided that some sample measurements are replicated across multiple batches. The 1-step method computes all calibration results iteratively from all measurements. It returns the most probable values for the sample compositions under the assumptions of a statistical model, making them the maximum likelihood predictors. It is less sensitive to measurement error on standards and enables use of some batches that do not include standards. In direct comparison of both real and simulated immunoblot data, the 1-step method consistently exhibited smaller errors than the conventional "2-step" method. These results suggest that the 1-step method is likely to be most useful for cases where experimenters want to analyze existing data that are missing some standard measurements and where experimenters want to extract the best results possible from their data. Open source software for both methods is available for download or on-line use.

Show MeSH
Calibrated experimental immunoblot data.This figure shows calibrated analyte amounts for 40 of our 230 samples that we analyzed with immunoblots. The others were qualitatively similar. Gray bars represent results from the 2-step method, red bars represent results from the 1-step method, and error bars represent standard error values. On average, there were 3.4 calibrated measurements for each sample with the 2-step method and 4.9 for the 1-step method. Note that the 1-step method results have smaller standard errors.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4764494&req=5

pone.0149575.g003: Calibrated experimental immunoblot data.This figure shows calibrated analyte amounts for 40 of our 230 samples that we analyzed with immunoblots. The others were qualitatively similar. Gray bars represent results from the 2-step method, red bars represent results from the 1-step method, and error bars represent standard error values. On average, there were 3.4 calibrated measurements for each sample with the 2-step method and 4.9 for the 1-step method. Note that the 1-step method results have smaller standard errors.

Mentions: We analyzed our experimental immunoblot data using both methods, of which a small portion of the results are shown in Fig 3. These data are scaled so that the standard (not shown in the figure) has an analyte amount of 1. As part of the analysis, we automatically removed all measurement results that were 4 or more standard deviations away from their expected values, which we deemed to be outliers, and then re-calibrated the remaining data repeatedly until there were no more outliers. This process showed that about 1% of our measurement results were outliers (for comparison, 0.003% would be expected to be more than 4 standard deviations away from the mean if errors were distributed perfectly normally). After all outliers were removed, the 1-step method enabled us to use all of the remaining measurements in the final analysis, whereas we needed to remove about 36% of them with the 2-step method. Results from both methods showed that the calibrated parameters for our data varied quite widely. Sample analyte amounts varied from 0 to about 2, where the standard was defined to have an amount of 1, and sensitivity parameters varied from 26 to about 700, in units of fluorescence units per amount of standard analyte. This latter 27-fold sensitivity range illustrates the wide variability of immunoblotting methods.


A Method and On-Line Tool for Maximum Likelihood Calibration of Immunoblots and Other Measurements That Are Quantified in Batches.

Andrews SS, Rutherford S - PLoS ONE (2016)

Calibrated experimental immunoblot data.This figure shows calibrated analyte amounts for 40 of our 230 samples that we analyzed with immunoblots. The others were qualitatively similar. Gray bars represent results from the 2-step method, red bars represent results from the 1-step method, and error bars represent standard error values. On average, there were 3.4 calibrated measurements for each sample with the 2-step method and 4.9 for the 1-step method. Note that the 1-step method results have smaller standard errors.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4764494&req=5

pone.0149575.g003: Calibrated experimental immunoblot data.This figure shows calibrated analyte amounts for 40 of our 230 samples that we analyzed with immunoblots. The others were qualitatively similar. Gray bars represent results from the 2-step method, red bars represent results from the 1-step method, and error bars represent standard error values. On average, there were 3.4 calibrated measurements for each sample with the 2-step method and 4.9 for the 1-step method. Note that the 1-step method results have smaller standard errors.
Mentions: We analyzed our experimental immunoblot data using both methods, of which a small portion of the results are shown in Fig 3. These data are scaled so that the standard (not shown in the figure) has an analyte amount of 1. As part of the analysis, we automatically removed all measurement results that were 4 or more standard deviations away from their expected values, which we deemed to be outliers, and then re-calibrated the remaining data repeatedly until there were no more outliers. This process showed that about 1% of our measurement results were outliers (for comparison, 0.003% would be expected to be more than 4 standard deviations away from the mean if errors were distributed perfectly normally). After all outliers were removed, the 1-step method enabled us to use all of the remaining measurements in the final analysis, whereas we needed to remove about 36% of them with the 2-step method. Results from both methods showed that the calibrated parameters for our data varied quite widely. Sample analyte amounts varied from 0 to about 2, where the standard was defined to have an amount of 1, and sensitivity parameters varied from 26 to about 700, in units of fluorescence units per amount of standard analyte. This latter 27-fold sensitivity range illustrates the wide variability of immunoblotting methods.

Bottom Line: The conventional approach has two steps: the experimenter deduces a conversion function using measurements on standards and then calibrates (or normalizes) measurements on unknown samples with this function.It returns the most probable values for the sample compositions under the assumptions of a statistical model, making them the maximum likelihood predictors.It is less sensitive to measurement error on standards and enables use of some batches that do not include standards.

View Article: PubMed Central - PubMed

Affiliation: Basic Sciences Division, Fred Hutchinson Cancer Research Center, Seattle, WA, United States of America.

ABSTRACT
Experimental measurements require calibration to transform measured signals into physically meaningful values. The conventional approach has two steps: the experimenter deduces a conversion function using measurements on standards and then calibrates (or normalizes) measurements on unknown samples with this function. The deduction of the conversion function from only the standard measurements causes the results to be quite sensitive to experimental noise. It also implies that any data collected without reliable standards must be discarded. Here we show that a "1-step calibration method" reduces these problems for the common situation in which samples are measured in batches, where a batch could be an immunoblot (Western blot), an enzyme-linked immunosorbent assay (ELISA), a sequence of spectra, or a microarray, provided that some sample measurements are replicated across multiple batches. The 1-step method computes all calibration results iteratively from all measurements. It returns the most probable values for the sample compositions under the assumptions of a statistical model, making them the maximum likelihood predictors. It is less sensitive to measurement error on standards and enables use of some batches that do not include standards. In direct comparison of both real and simulated immunoblot data, the 1-step method consistently exhibited smaller errors than the conventional "2-step" method. These results suggest that the 1-step method is likely to be most useful for cases where experimenters want to analyze existing data that are missing some standard measurements and where experimenters want to extract the best results possible from their data. Open source software for both methods is available for download or on-line use.

Show MeSH