Standard machine learning approaches outperform deep representation learning on phenotype prediction from transcriptomics data

Table 3 TCGA overall survival (OS) tasks

TCGA OS tasks
Project	Disease	Label	Label type	Group	Samples
CESC	cervical squamous cell carcinoma	OS	survival	train	304
COAD	colon adenocarcinoma	OS	survival	train	455
ESCA	esophageal carcinoma	OS	survival	train	184
KIRP	kidney papillary cell carcinoma	OS	survival	train	289
LUAD	lung adenocarcinoma	OS	survival	train	507
OV	ovarian cancer	OS	survival	train	420
PAAD	pancreatic adenocarcinoma	OS	survival	train	178
SARC	sarcoma	OS	survival	train	259
STAD	stomach adenocarcinoma	OS	survival	train	409
UCEC	uterine corpus endometrial carcinoma	OS	survival	train	540
HNSC	head-neck squamous cell carcinoma	OS	survival	validate	501
BLCA	bladder urothelial carcinoma	OS	survival	test	407
LUSC	lung squamous cell carcinoma	OS	survival	test	495

The 13 overall survival tasks derived from TCGA used to train supervised models and validate the unsupervised embeddings. The project names correspond to those in Fig. 2

ISSN: 1471-2105