Table 1. Overview of compartmental assignment results.

Compound (cpd.) class	unexplained			specific			dominant			enriched			shared/shared*				total
	(cpl)	(cyt)	(vac)	cpl	cyt	vac	cpl	cyt	vac	cpl	cyt	vac	cpl <> cyt	cyt <> vac	cpl <> vac	other
primary cpd.	8	24	9	7	12	9	11	28	15	4	10	6	15	24	1	20	203
	41 (20.2%)			28 (13.8%)			54 (26.6%)			20 (9.9%)			60 (29.6%)
lipophilic cpd.	146	342	24	326	288	0	284	756	3	49	125	0	270	102	1	88	2804
	512 (18.3%)			614 (21.9%)			1043 (37.2%)			174 (6.2%)			461 (16.4%)
secondary cpd.	20	119	30	10	161	158	39	44	76	14	14	21	48	55	22	79	910
	169 (18.6%)			329 (36.2%)			159 (17.5%)			49 (5.4%)			204 (22.4%)
others (+)	0	2	0	1	0	0	0	0	0	0	0	0	1	0	0	1	5
	2 (40%)			1 (20%)			0 (0%)			0 (0%)			2 (40%)
total	174	487	63	344	461	167	334	828	94	67	149	27	334	181	24	188	3922
	724 (18.5%)			972 (24.8%)			1256 (32%)			243 (6.2%)			727 (18.5%)

Classification of analytes into classes is based on the BFA-estimated subcellular distributions (Data S4) derived from three independent gradients using a classification tree (Figure 6). Venn diagrams are depicted in Figure S3. Analytes with insufficiently explained (unexplained) distributions using the selected compartment-specific markers are classified by the kNN algorithm using the three nearest neighbor (k = 3) compartment-specific markers.