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

Table 4 TCGA progression-free interval (PFI) tasks

TCGA PFI tasks
Project	Disease	Label	Label type	Group	Samples
CESC	cervical squamous cell carcinoma	PFI	survival	train	304
COAD	colon adenocarcinoma	PFI	survival	train	455
ESCA	esophageal carcinoma	PFI	survival	train	184
KIRP	kidney papillary cell carcinoma	PFI	survival	train	288
LUAD	lung adenocarcinoma	PFI	survival	train	507
OV	ovarian cancer	PFI	survival	train	420
PAAD	pancreatic adenocarcinoma	PFI	survival	train	178
SARC	sarcoma	PFI	survival	train	259
STAD	stomach adenocarcinoma	PFI	survival	train	411
UCEC	uterine corpus endometrial carcinoma	PFI	survival	train	540
HNSC	head-neck squamous cell carcinoma	PFI	survival	validate	501
BLCA	bladder urothelial carcinoma	PFI	survival	test	408
LUSC	lung squamous cell carcinoma	PFI	survival	test	496

The 13 progression-free surivival 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