Related Experiment Video
Updated: Dec 5, 2025
06:52
Discovery of Driver Genes in Colorectal HT29-derived Cancer Stem-Like Tumorspheres
Published on: July 22, 2020
6.8K
Machine learning application identifies novel gene signatures from transcriptomic data of spontaneous canine
Nuojin Cheng1, Ashley J Schulte2, Fadil Santosa3
1School of Mathematics, College of Science and Engineering at the University of Minnesota, Minneapolis, MN, USA.
Briefings in Bioinformatics
|October 20, 2020
Summary
Machine learning accurately diagnoses canine hemangiosarcoma using transcriptomic data. This approach identifies novel gene signatures for vascular malignancy, improving diagnostic potential.
Area of Science:
- Veterinary Oncology
- Bioinformatics
- Genomics
Background:
- Angiosarcomas are rare, aggressive vascular soft-tissue sarcomas with poor outcomes.
- Canine hemangiosarcoma shares features with human angiosarcoma, serving as a model.
- Histological diagnosis can be challenging due to sample limitations.
Purpose of the Study:
- To evaluate machine learning models for diagnosing canine hemangiosarcoma using transcriptomic data.
- To identify novel gene signatures for improved diagnostic accuracy.
Main Methods:
- Applied machine learning (Extra Trees, Random Forest) to next-generation transcriptomic data from 76 canine hemangiosarcoma and 10 nonmalignant samples.
- Utilized 10-fold cross-validation and feature selection methods (mutual information, Monte Carlo).
Main Results:
- Extra Trees and Random Forest models demonstrated high classification accuracy for hemangiosarcoma.
- Identified novel gene signatures with potential as reliable diagnostic markers.
- Confirmed trainability of high-throughput sequencing data for machine learning applications.
Conclusions:
- Machine learning applied to transcriptomic data is a viable tool for canine hemangiosarcoma diagnosis.
- The identified gene signatures offer insights into vascular malignancy and potential diagnostic applications.
