. 2020 Nov 5;11:5595. doi: 10.1038/s41467-020-19354-z

Table 6.

Description of datasets.

MS instruments	Datasets	Classes	# spectra	# samples	Mass ranges	# features
Target domain data
Synapt G2-S Q-TOF (Waters, SpiderMass)	Canine sarcoma	Healthy	482	8	100–1600 Da	15,000
		Myxosarcoma	60	1
		Fibrosarcoma	404	6
		Hemangiopericytoma	134	2
		Malignant peripheral nerve tumor	60	1
		Osteosarcoma	339	5
		Undifferentiated pleomorphic sarcoma	376	5
		Rhabdomyosarcoma	66	1
		Splenic fibrohistiocytic nodules	63	1
		Histiocytic sarcoma	105	1
		Soft tissue sarcoma	69	1
		Gastrointestinal stromal sarcoma	70	1
		Total	2228	33 biopsies
Synapt G2-S Q-TOF (Waters, SpiderMass)	Microorganisms	Staphylococcus aureus	26	1	100–2000 Da	19,000^a
		E. coli D31	26	1		15,000^b
		Pseudomonas aeruginosa	24	1
		Enterococcus faecalis	18	1
		Candida albicans	23	1
		Total	117	5 colonies
PBSII SELDI-TOF	Human ovary 2	Healthy	91		700–12,000 Da	7084
		Cancer	162	³⁷
		Total	253
Source domain data
Rapiflex MALDI-TOF (Bruker)	Rat brain	Gray matter	4635	A single section	300–1300 Da	19,000^a
		White matter	5465			15,000^b
		Total	10100			7084^c
Synapt G2-S Q-TOF (Waters, SpiderMass)	Beef liver	Positive mode	1372	10	100–1600 Da	15,000
		Negative mode	1265	10
		Total	2637	20 samples
Hybrid Quadrupole (QSTAR pulsar I)	Human ovary 1	Healthy	95		1–20,000 Da	7084
		Cancer	121	³⁷
		Total	216

^aNumber of features used for microorganisms transfer learning.

^bNumber of features used for canine sarcoma transfer and cumulative learning.

^cNumber of features used for ovarian transfer and cumulative learning.