To test the reliability of our software OCAS, we compare the results obtained using OCAS and 3 other software, namely the widely used Automated Mass spectral Deconvolution and Identification System (AMDIS) and 2 open source software.
Firstly, a controlled sample consisting of equal volumes of:
1) Methanol, 2) Acetonitrile, 3) Methylene Chloride, and 4) 1, 4-Dioxane
were run through an Agilent GC-MS instrument. The GC column (HP5-MS) temperature is fixed at 240 oC. The experiment was engineered such that the TIC of these 4 components would show that their peaks co-eluted to form a single, symmetrical peak. We then processed the data obtained using all the software and compared the results of each software in the table below.
- More Accurate Results: OCAS identified all 4 components correctly from the controlled sample while the 2 open source software did not manage to do so.
- More Precise Results: OCAS returned precisely the 4 components in the mixture while all other software returned much more possible chemical components and users had to sieve through the information to look for the correct ones.
- Concentration Information: OCAS was able to calculate and give a good estimate of the concentration of each component in the mixture while all other software were unable to do so. OCAS could even visually display the deconvoluted peaks in the same graph as the original peak to provide users with a better understanding of the co-elution.
- Ease of Use: 4 clicks of the mouse button are all it takes for OCAS to produce the results while the other software required users to input many parameters before the results could be obtained.
Quantitative Analyses: In another analysis of the same mixture, we used the normal GC method to separate these four components to calculate the actual TIC concentration of each pure component with the area normalization method. The four components’ retention time and area ratio was calculated. We also calculated the ratio of estimated TIC concentration for each estimated spectrum. The results are listed in table below.
Table. Comparison of retention time and area percentage between real and EstConcs
These results demonstrate the following:
1) The eluting sequence of components inside the mixture sample corresponded to the real result.
2) Retention time of each reconstructed Spectrum corresponded to that of real pure compounds.
3) The area percentages matched well between the real and reconstructed concentration.
The quantitative result of each reconstructed Concentration inside the co-eluted peak was accurate. Since the area of each reconstructed concentration now links to the real TIC area, it is shown to produce results compatible with those of the current standard quantitative methods. As a result, external and internal standard methods support that our algorithm enables an accurate quantitative analysis of components inside co-eluted peaks.