IonVision Data Structure
And Visualization in R

Loading sample data

Let’s begin by looking at the data structure we have in a typical dataset. This example set consists of seven measurements. Two of ethanol, two of 2-propanol, and three background measurements taken before sampling, after ethanol, and after 2-propanol. The table below presents a printout of the first columns of the dataset, containing measurement comments, environmental parameters etc. The loading functions used always ensure that the data is ordered by start time, making the oldest result the first one in the table.

As seen in the table, we load some select environmental parameters, along with the comment, start time, and all data from the measurement files. Flows are measured from the sample (the input of the device) and the sensor (the flow through the DMS sensor, which includes the sample flow and the clean circulating flow inside the device). Temperature and humidity are measured from both of the flows and additionally from inside the device to give a reference point to the ambient environment. Maximum, minimum and average values are collected from all these parameters.

The environmental parameters might be of special interest in long measurement series, or when the repeatability of the system or a measurement is tested. DMS is very dependent on the absolute humidity of the sample. Lowering the humidity increases the sensitivity of the device. This means that changing environmental conditions may present challenges when sampling is done from ambient conditions. The system also allows the user to set upper and lower limits for these parameters, and any ”errors” from these limits are recorded in the json file as well. This provides a consistent way to exclude measurements where certain conditions are not met.

Comments and annotation

When used in a consistent manner, the comments can be used to annotate the measurements for data analysis. Another possibility is to use external annotations from a .csv file. Let’s say we needed the comments to be identical for repeat measurements of background, ethanol, and 2-propanol. As seen in the first table, we have some minor inconsistencies in the commenting. With such a small set, we could edit the comments manually in R or from the .json files, or employ the external annotation.

For example’s sake, let’s load the external annotations found in the sample data folder. The below table shows a comparison between the original comments and the example annotation file.

In cases where several samples need to be averaged (like in the below section), consistent comments make looping through the data easier. It is also possible to use, for example, sample IDs as the IonVision comments depending on your use case.

Data visualization

Now that we have the data loaded and properly annotated, we can plot the average responses for our three classes. IonVision data consists of two spectra, one from the positive ion side and one from the negative ion side. Number of data points of both sides is determined by the Usv and Ucv vectors. Usv is the steps taken by the separation voltage, and Ucv by the compensation voltage. Typically, the Usv is plotted as the y-axis and Ucv as the x-axis. The color scale is the signal intensity in pA. The negative ions are recorded as a negative pA reading, but for these images the absolute values are shown to keep the color scale consistent.