Lung cancer type classification using differentiator genes
| Autor: | Melford John, Sterling Ramroach, Ajay Joshi |
|---|---|
| Rok vydání: | 2020 |
| Předmět: |
0301 basic medicine
Oncology medicine.medical_specialty business.industry Cancer type Early detection Disease medicine.disease 03 medical and health sciences Differentiator 030104 developmental biology 0302 clinical medicine Expression data 030220 oncology & carcinogenesis Internal medicine Genetics medicine Lung cancer business Survival rate Gene |
| Zdroj: | Gene Reports. 19:100647 |
| ISSN: | 2452-0144 |
| Popis: | Almost one million patients are diagnosed with some form of Lung cancer every year. It is one of the deadliest malignant tumours worldwide and early detection is known to significantly improve a patient's survival. There is a need for early diagnosis, accurate histological subtype classification, and targeted treatment. Using 1019 lung cancer samples, each containing expression data for 18,015 genes, we trained a neural network to classify each sample. We also used the Gini coefficient of each gene to extract a subset of differentiator genes. After 10-fold cross validation and using only the most important differentiator genes (DEFB108B, DEFB136, KRTAP19.4, OR10H4, SUN5, and TXNDC8), our network scored an average accuracy of 99.47% ± 0.57, at least 5% higher than previous works, in classifying cancer type. Patients with alterations to the differentiator genes showed a lower survival rate with less months being disease/progression free. These results support the existence of potential drug targets in the set of differentiator genes. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|
Full Text from ScienceDirect