Data Management And Sharing System of China National Clinical Research Center for Neurological Diseases
Predicting chromosome 1p/19q codeletion by RNA expression profile: a comparison of current prediction models
First author：Zhi‐liang Wang
Corresponding author：Jiang Tao
essential_information
- Person in charge：Zhi‐liang Wang
- Data Sources： OTHERS
- Number of papers：OTHERS20190785
- Published time：2019-02-02
2019-02-02
Published articles
Journal：Aging
Publication type: Research Paper
IF(IF): 5.515
PMID: 30710490 Link To Pubmed Full text Download
Background And Purpose
Method
We included 692 samples with 1p/19q status information from TCGA cohort as training set and 222 samples with 1p/19q status information from REMBRANDT cohort as validation set. We reviewed and compared five tools: TSPairs, GSVA, PAM, Caret, smoother, with respect to their accuracy, sensitivity and specificity.
Conclusion
Our independent assessment of five tools revealed that smoother method was selected as the most stable and accurate method in predicting 1p/19q status. This method could be regarded as a potential alternative method for clinical practice in future.
Main results
In TCGA cohort, the GSVA method showed the highest accuracy (98.4%) in predicting 1p/19q status (sensitivity=95.5%, specificity=99.6%) and smoother method showed the second-highest accuracy (accuracy=97.8%, sensitivity=96.4%, specificity=98.3%). While in REMBRANDT cohort, smoother method exhibited the highest accuracy (98.6%) (sensitivity= 96.7%, specificity=98.9%) in 1p/19q status prediction.