PheSeq, a Bayesian deep learning model to enhance and interpret the gene-disease association studies

Table 5 Comparison of different data fusion methods on gene-disease associations

Method	Data modality	Strategy^a	Interpretation	Count^b	Hit^c
Alzheimers’ disease
Lee 22 [78]	GWAS + eQTL	Early	/	12	8
DeepGAMI [81]	Gene expression + GRN +eQTL	Early	Enrichment analysis	102	2
scGRNom [82]	GWAS + Hi-C	Intermediate	Enrichment analysis	146	43
GDAMDB [32]	GWAS \(+\) literature	Late	/	149	72
PheSeq	GWAS \(+\) literature \(+\) PPI	Late	Phenotype and literature	1024	151
Breast cancer
Kim 20 [79]	CNA + gene expression + methylations + clinical info	Intermediate	/	36	5
Ahn 14 [83]	Transcriptome + pathway	Early	/	50	2
GLRP [84]	Gene expression + PPI	Early	Interpretable neural network + pathway analysis	167	5
IMNA [85]	GWAS + eQTL	Early	Literature	391	24
PheSeq	Transcriptome \(+\) literature \(+\) PPI	Late	Phenotype and literature	818	159
Lung cancer
Zhang 20 [80]	Methylation + gene expression	Early	/	23	2
Gogleva 22 [86]	CRISPR + knowledge graph + literature	Intermediate	Interpretable recommendation system	31	8
CTpathway [87]	Gene expression + transcriptome + pathway + PPI	Intermediate	Pathway analysis	59	12
ECMarker [88]	Gene expression + clinical phenotype	Early	Interpretable neural network	500	19
PheSeq	Methylation \(+\) literature \(+\) PPI	Late	Phenotype and literature	566	342

^aData fusion strategy. Early for data-level, intermediate for joint-level, and late for decision-level
^bCount of significant genes
^cHits of significant genes in DISEASES

ISSN: 1756-994X