Pure ion profiles
In X-HRMS, each scan is an image of the composition of the mixture leaving the separation device. A pure ion profile (PIP) is obtained with centroid dataset and consists in regrouping along every scan all data points that are related to one and only one ion. This is usually done by assuming that consecutive points belong to the same ions if their accurate masses do not differ by more than a set threshold. Pure ion profiles algorithms will try to extract from a dataset the profiles of all ions resulting from the separation of the compounds from the matrix by the separation technique and their ionisation and separation by the mass spectrometer. Once PIP obtained, figures of merits (e.g. peak area, centre, variances...) can easily be calculated and summarised inside a table often known as the peak list.
In the toolbox, the PeakList object deals with extraction of pure ion profiles, the peak list as well as different representation associated with it. It is an independent object that is saved in the .fin folder. To object is initialise using the following command:
myPeakList = PeakList(dtsIn, thrInt, deltmz, minPts)
where dtsIn is a centroid Dataset (i.e. myFinnee.Datasets{4}) and deltmz is the m/z threshold that will determine if two points in successive scans belong to the same PIP. Typically deltmz should be bellow 0.1 m/z. Only PIPs with more than minPts and with, at least, one point of intensity higher than thrInt will be recorded. The value for minPts depends on the MS scan rate and profile peak width, but should not be bellow 5. The value for thrInt should be equal or greater than the threshold used for noise removal. The PIP algorithms developed for PeakList should work for any centroid dataset. However, it was designed for datasets that were corrected for baseline drift and background noise. Best results will be obtained if those steps were performed. For example with the CE-TOFMS dataset, we used
myPeakList = PeakList(myFinnee.Datasets{5}, 100, 0.1, 5);
where Datasets{5} was obtained as follow:
- Noise reduction. myFinnee = myFinnee.filterDataset(1, 'RemoveNoise:5:5:100');
- Baseline correction. myFinnee = myFinnee.BaselineCorrection(2);
- Noise reduction. myFinnee = myFinnee.filterDataset(3, 'RemoveNoise:5:5:100');
- Centroidisation. myFinnee = myFinnee.doCentroid(4, 'LocalMax:2:0');
PeakList objects contain the list of all PIPS that were constructed from the dataset as well as associated figures of merits (FOM).
The base peak profile (BPP) and total ion profile (TIP) in myPeakList are constructed using only the data points that are arranged in PIPs. This allows verifying the performance of the algorithms.
LstPIP contains all individual PIPs. In this example, 912 PIPs were extracted from the dataset. Each PIP can be easily visualised using
myPeakList.LstPIP{Id}.plot
where Id is the index to one of the PIP.
The plot method superposes a profile-like representation with the position in the m/z axis of all points that constitute the PIP. While ideally, the PIP algorithms should only select one data points per scan for a given PIP, this is not often the case. Additional points can be present either due to background noise or because ions with very close m/z are not baseline separated. The incidence of such events can be decreased by using smaller deltmz. However, this can also result in incomplete profiles.
The right panel in the figure summarised key figures of merits. Those are the intensity at PIP apex, the time at PIP apex, the peak area, the peak centre, the peak variance, the average m/z values of all points with the standard deviation between brackets and the accurate mass. Peak area, centre and variance are calculated using statistical moments. The accurate mass is the sum of the product of the intensities and m/z values of all points divided by the sum of intensities.
FOM is a structure that summarised the figure of merits of all PIPs. myPeakList.FOM possess two fields, Headings and Data, where Data in a 10xn array. FOM.Data could be used to filter the PIPs, compare dataset... Whithin the frame of this work, we used PeakList to develop two new representations facilitating a comprehensive analysis of X-HRMS datasets. Those are the centergram and clusters plot.
Up : Pure ion profile and peak list
Next : Centergram representation
Previous : Pure ion profiles and peak list
Related to: