Statistical approaches to analyzing large numbers of spectra often involve data reduction methods like spectral binning to reduce the number of variables that must be considered. Spectral binning divides spectra into a series of regions, or bins; subsequent analysis involves the integrated area of these bins instead of the raw spectral data. Profiler lets you perform spectral binning on collections of spectra, exporting the results to an excel document (.xlsx) and to tab-delimited text (.tsv) that is easily read by spreadsheet programs and statistical analysis software packages.
| Menu Location | Tools > Spectral Binning... |
| Icon |
 |
How do I bin spectra?
-
Start the Spectral Binning wizard.
-
Choose the source spectra (
.cnx) for binning, and specify a target folder for the resulting binned data. You can specify a series of spectrum files or one or more folders containing spectrum files as the source.If you would like to repeat a binning session that you have previously saved, you can specify a binning configuration file (
BinningConfiguration.xml) to import.A binning configuration file is automatically created each time you calculate binning results.
-
Choose a binning method to apply to the specified spectra. You can use a defined bin size in ppm or a defined number of bins. You can also use bins based on the transform windows of the compounds in a Compound Set ("targeted binning").
If you would like to reuse a predefined set of bins, you can specify a bin definition file (
BinDefinitions.tsv) to load. -
Choose regions of the spectrum to exclude from binning, if desired. You can specify any regions of the spectrum using start and end values in ppm.
-
Choose a binning target. You can bin spectra based on the sum line (standard binning) or the subtraction line (residual binning).
-
Choose a normalization method. You can normalize the bins relative to the total area under each spectrum line or to units of standardized area (sa). Note that 1 sa is the area under a theoretical DSS methyl peak at 0.50 mM in each spectrum.
Tips and Tricks
-
Binning can take a long time, especially on older computers or particularly large datasets. If you have never tried binning before, try a small dataset of five or ten spectra first, before moving on to larger datasets. The time needed for the smaller dataset will help you to more accurately estimate the time needed to complete a binning session.
-
If you choose one or more folders as the source, binning includes spectra both directly in the specified folders and in subfolders of the specified folders.
-
Binning based on number of bins divides the spectrum into the specified number of bins after excluding the regions that you have defined, if any.
-
To do residual binning, each of the source spectra must have been analyzed in Profiler and must contain at least one compound with a defined concentration.
-
Excluding regions from a binning dataset can help you to remove the influence of interfering or confounding signals, such as solvent peaks.
-
Normalization based on total area gives the area of a particular bin as a fraction of the total area of all bins, except area in any of the excluded regions that you may have defined.
-
Spectral binning is subject to several numerical limits:
-
Each binning session can include a maximum of 5000 spectra.
-
The range that you define when using number of bins or size of bins as a binning method must be at least 1 ppm wide, and can be as large as 1998 ppm (-999 to 999 ppm).
-
You can specify a maximum of 100000 (one hundred thousand) bins or a minimum of 1.
-
You can specify a minimum bin width of 0.00001 ppm and a maximum bin width equal to the width of the range that you have chosen. For example, if you are binning across a range of 2 to 10 ppm, your maximum bin width is 8 ppm.
-
Related Topics