Machine Learning Supports Long Noncoding RNAs as Expression Markers for Endometrial Carcinoma.
Ana Carolina MelloMartiela FreitasLaura CoutinhoTiago FalconUrsula da Silveira MattePublished in: BioMed research international (2020)
Uterine corpus endometrial carcinoma (UCEC) is the second most common type of gynecological tumor. Several research studies have recently shown the potential of different ncRNAs as biomarkers for prognostics and diagnosis in different types of cancers, including UCEC. Thus, we hypothesized that long noncoding RNAs (lncRNAs) could serve as efficient factors to discriminate solid primary (TP) and normal adjacent (NT) tissues in UCEC with high accuracy. We performed an in silico differential expression analysis comparing TP and NT from a set of samples downloaded from the Cancer Genome Atlas (TCGA) database, targeting highly differentially expressed lncRNAs that could potentially serve as gene expression markers. All analyses were performed in R software. The receiver operator characteristics (ROC) analyses and both supervised and unsupervised machine learning indicated a set of 14 lncRNAs that may serve as biomarkers for UCEC. Functions and putative pathways were assessed through a coexpression network and target enrichment analysis.
Keyphrases
- machine learning
- network analysis
- gene expression
- genome wide identification
- artificial intelligence
- big data
- papillary thyroid
- poor prognosis
- genome wide analysis
- endometrial cancer
- dna methylation
- deep learning
- single cell
- molecular docking
- squamous cell
- childhood cancer
- lymph node metastasis
- adverse drug
- long non coding rna