Calibration and Concentration Determination¶

Calibration Object¶

The calibration class Calibrate is defined in the gcmstools.calibration module. Before you start, you must have access to an HDF storage file that contains all of the data to be processed. By default, the Calibrate instance creation attempts to open a file “data.h5” in the current folder. If this is not the name of your HDF file, an alternate name can also be passed in on construction to open a different HDF file.

In : from gcmstools.calibration import Calibrate

In : cal = Calibrate() # Opens 'data.h5' by default

# Equivalent to above, but using a different file name, don't do both
In : cal = Calibrate('other.h5')

Closing the HDF File¶

In general, you will want to close the HDF file when you’re done. This is not necessary, but it does ensure that the file gets properly compressed, which saves some disk space. If you don’t do this, though, it won’t hurt anything.

In : cal.close() # Only do this when you're done

Calibration Information File¶

In order to calibrate your GCMS data, you must first create a csv file containing all of the relevant calibration information. Again, the structure of this file is very important, so an example, “calibration.csv”, is contained with the sample data.

In : from gcmstools.general import get_sample_data

In : get_sample_data("calibration.csv")

The first row in this csv file is critical, and it must look like this:

Compound,File,Concentration,Standard,Standard Conc

Each row after this describes a set of calibration information that you’d like to use. The columns of this file are as follows:

Compound: The name of the compound that you are calibrating. This must correspond to one of the compound names (case-sensitive) used when referencing and fitting the GCMS file.
File: This is the name of the data set that was collected at a particular concentration of Compound. Again, this filename can be the full filename (with or without the path) or the simplified name. See the files attribute section of the GcmsStore docs for more info.
Concentration: This is the known concentration of the Compound. This should only be a number: do not include units. All of the concentrations should be in the same units, and keep in mind that all calibration and concentration data will then be in that same unit of measurement.
Standard and Standard Conc: If there is an internal standard used in this file, you should provide the name and concentration in these columns. Again, the standard name should have been defined when referencing your GCMS data set, and do not include units with the concentration. Make sure your concentration units are the same as the reference compound to avoid confusion.

You can add extra columns to this table without penalty, in case you need to add additional information to this table. You can also comment out lines by starting a line with a # character. This is useful if you want to ignore a bad data point without completely removing the line from the calibration file.

Generate Calibration Curves¶

The calibration curves can be generated using the curvegen method of the Calibrate object. This function must be called with the name of your calibration file. In this example, that filename is “calibration.csv”.

In : cal.curvegen('calibration.csv')
Calibrating: benzene
Calibrating: phenol
...

In :

This process creates two new tables as attributes to your calibration object, calinput and calibration. The former table is simply your input csv information with columns appended for the concentrations (“conc”) and integrals (“integral”) used for generating the calibration curve. If no internal standard is defined, then “conc” will be the same as the compound concentration you used in the input file. If an internal standard was defined, then “conc” and “integral” will be these values divided by the corresponding internal standard values. These tables also stored in the HDF file as well, if you want to check them at a later date.

The calibration table contains all of the newly created calibration curve information, such as slope, intercept, r value, etc.

In : cal.calibration
Out:
               Start  Stop  Standard         slope      intercept         r  \
Compound
benzene          2.9   3.5       NaN  38629.931565 -367129.586850  0.998767
phenol          14.6  15.1       NaN  30248.192619   65329.897933  0.999136
...

                      p       stderr
Compound
benzene        0.000052  1108.344872
phenol         0.000030   726.257380
...

Plotting Calibrations¶

By default, no plots are generated for these calibrations. There are, however, a couple of functions that automatically plot some of the calibration data.

cal.curvegen('calibration.csv', calfolder='cal', picts=True) : This invocation will auto generate pictures for all of the calibration compounds and place them in a folder defined by the keyword argument calfolder. This argument is optional, if you don’t mind the default folder name of “cal”. Be careful! This folder and its contents will be deleted before generating new plots, so if this folder exists, make sure it is clear of important data.
cal.curveplot('benzene') : This method will generate a plot of the benzene calibration information and save it to the current folder. There are several keyword arguments to this function:
- folder='.' : This sets the folder where the picture will be saved. By default it is the current directory.
- show=False : Change this value to True if you want an interactive plot window to be displayed. Default is False.
- save=True : Save the calibration plot to the folder.
If both save and show are set to False, nothing will happen.

Of course, this function must be done after a call to curvegen. But it can be used to look at calibration data from an previously processed HDF file without rerunning the calibration.

Determine Sample Concentrations¶

Generating calibration curves does not automatically process the other data files. In order to determine concentrations for all of the remaining data in the HDF file, use the datagen method of the Calibrate object.

In : cal.datagen()
Processing: datasample1.CDF
Processing: otherdata1.CDF
Processing: otherdata2.CDF
...

After processing, another data table attribute (datacal) is created and saved to the HDF file.

In : cal.datacal
Out:
                              benzene       phenol   ...
name
datasample1               4239.070627    58.336917   ...
otherdata1                5475.778519    20.401981   ...
otherdata2                4355.094930    19.171877   ...
...

Note

Again, the data ARE NOT automatically integrated after generating calibration curves. If you change your calibration information by re-runing curvegen, you must re-run datagen to apply these changes to the other data sets contained in the HDF file.

Plotting Integrals with Concentrations¶

By default, no plots are generated for the integrals. If you’d like to see plots of the integrals, there are a couple of methods.

cal.datagen(datafolder='data', picts=True) : This method will auto generate pictures for all of the calibration compounds and place them in a folder defined by the keyword argument datafolder. This argument is optional, if you don’t mind the default folder name of “data”. Be careful! This will delete this folder before generating new plots, so if this folder exists, make sure it is clear of important data.
cal.dataplot('benzene', 'datasample1') : This method will generate a plot of the benzene integral for ‘datasample1’ and save it to the current folder. There are several keyword arguments to this function:
- folder='.' : This sets the folder where the picture will be saved. By default it is the current directory.
- show=False : Change this value to True if you want an interactive plot window to be displayed. Default is False.
- save=True : Save the calibration plot to the folder.
If both save and show are set to False, nothing will happen.

Of course, this function call can only be done after a call to datagen, but it can be used to look at calibration data from an previously processed HDF file without rerunning the calibration and data integration functions.