Fig. 8: Post–projection calibration improves unknown identities annotation. | Communications Chemistry

Fig. 8: Post–projection calibration improves unknown identities annotation.

From: Generic and accurate prediction of retention times in liquid chromatography by post–projection calibration

Fig. 8

There are 2935 putative candidates for a total of 112 unknown identities and each identity has more than 3 putative candidates. Panels a–j show the improvement of ranking putative candidates. Panels d–g show the improvement of filtering putative candidates. Examples are given of some molecules ranking relatively poorly in OCMs 19–24 (a), some molecules ranking relatively poorly in OCMs 25–30 (b), and some molecules ranking almost uniformly across all OCMs (c) before calibration. The ranking of these molecules reached similar levels among all OCMs after calibration (a–c). Panels h, i show the differences in the number of correct candidates is exactly the Nth top candidate among 30 OCMs before (h) and after calibration (i). An example is given of using post–projection calibration method to improve the correct candidate ranking (j). Panel d shows the filtering error threshold for 30 OCMs. Panel e shows the true negative rate (TNR). Panel f shows the true positive rate (TPR). Panel g shows the accuracy of filtering putative candidates by RT error threshold.

Back to article page