Table 2.

Summary of the performance of different algorithms for complex samples

Sample	Identified and annotated metabolites	Tool	Annotated metabolites		Adducts/mass fragments	Annotated features
			Multiple adducts	Single adduct
Retina IRS2 KO (+ ionization)	20	CliqueMS	15	—	50	95
		CAMERA	8	—	25	45
Retina IRS2 KO (− ionization)	18	CliqueMS	6	—	16	35
		CAMERA	5	—	14	33
MTBLS103 HILIC	6 (78)a	CliqueMS	5/6/56b	—	18/26/213b	44/72/318b
	6	CAMERA	3	—	13	21
		xMSannotator	1	4	10	10
		MS-FLO	1	—	2	3
MTBLS103 C18	9 (162)a	CliqueMS	6/8/104b	—	17/29/304b	46/66/524b
	9	CAMERA	3	—	11	20
		xMSannotator	3	6	13	13
		MS-FLO	0	—	0	0

Note: For single spectrum samples (Retina IRS2 KO in positive and negative ionization mode), we report results for CAMERA and CliqueMS. For the datasets in MTBLS103 (Samino et al., 2015), we report results for the chromatographic column operating in two different conditions: RP-C18 and HILIC. For the MTBLS103 datasets, we show results for CliqueMS, CAMERA, xMSAnnotator and MS-FLO. The multiple adduct and single adduct columns indicate the number of correctly annotated metabolites through the identification of at least two adducts with the same parental neutral mass, and the number of metabolites annotated through the annotation of a single adduct [annotated single adducts are assigned to (M + H)+ by xMSannotator].

^aCliqueMS analyzes individual samples, therefore in parenthesis we show the total number of annotated metabolites in all samples.

^bBecause CliqueMS produces an individual annotation for each sample (13 for HILIC and 18 for RP-C18), we report three results $r_1/r_2/r_3$: r₁ shows the number of unique metabolites/adducts/features that are correctly annotated in $\ge 50\%$ of the samples; r₂ shows the number of unique metabolites/adducts/features which are correctly annotated in at least one sample and r₃ shows the aggregate numbers over samples.