-
Categories
-
Pharmaceutical Intermediates
-
Active Pharmaceutical Ingredients
-
Food Additives
- Industrial Coatings
- Agrochemicals
- Dyes and Pigments
- Surfactant
- Flavors and Fragrances
- Chemical Reagents
- Catalyst and Auxiliary
- Natural Products
- Inorganic Chemistry
-
Organic Chemistry
-
Biochemical Engineering
- Analytical Chemistry
- Cosmetic Ingredient
-
Pharmaceutical Intermediates
Promotion
ECHEMI Mall
Wholesale
Weekly Price
Exhibition
News
-
Trade Service
On December 28, 2022, Li Hong's research group of the Shanghai Institute of Nutrition and Health, Chinese Academy of Sciences published an online report entitled "A systematic assessment of deep learning methods for drug response prediction: from in vitro to clinical" in the international academic journal Briefings in Bioinformatics applications"
.
This paper systematically evaluates the performance of deep learning algorithms for cancer susceptibility prediction from multiple perspectives, provides guidance for users to choose appropriate prediction models according to their own needs and data characteristics, and guides the construction
of new computational models.
.
This paper systematically evaluates the performance of deep learning algorithms for cancer susceptibility prediction from multiple perspectives, provides guidance for users to choose appropriate prediction models according to their own needs and data characteristics, and guides the construction
of new computational models.
Predicting the killing effect of drugs on tumors based on molecular omics is an important direction
in personalized cancer treatment.
Recent research results show that deep learning models can improve prediction performance
compared to classical machine learning models.
However, there is currently a lack of systematic comparison of different deep learning methods, especially model transfer capabilities
from preclinical models to clinical data.
The researchers evaluated the performance of six representative deep learning methods for susceptibility prediction in multiple application scenarios using nine evaluation metrics, including overall prediction accuracy, predictability at the individual drug level, potential correlations in predictive performance, and the ability of cell line models to migrate to clinical patients
.
The results show that most methods have good predictive performance in cell line datasets, among which the dual-graph neural network algorithm based on protein-protein association has stronger ability to capture omics features of tumor cell lines and better accuracy
.
Due to the differences between cell lines and patient tumor samples, the performance indicators of models trained on cell lines will be reduced to varying degrees when applied to patients, but several single-drug models can still achieve reliable prediction results
on some drugs.
in personalized cancer treatment.
Recent research results show that deep learning models can improve prediction performance
compared to classical machine learning models.
However, there is currently a lack of systematic comparison of different deep learning methods, especially model transfer capabilities
from preclinical models to clinical data.
The researchers evaluated the performance of six representative deep learning methods for susceptibility prediction in multiple application scenarios using nine evaluation metrics, including overall prediction accuracy, predictability at the individual drug level, potential correlations in predictive performance, and the ability of cell line models to migrate to clinical patients
.
The results show that most methods have good predictive performance in cell line datasets, among which the dual-graph neural network algorithm based on protein-protein association has stronger ability to capture omics features of tumor cell lines and better accuracy
.
Due to the differences between cell lines and patient tumor samples, the performance indicators of models trained on cell lines will be reduced to varying degrees when applied to patients, but several single-drug models can still achieve reliable prediction results
on some drugs.
Bihan Shen, a doctoral student at the Shanghai Institute of Nutrition and Health, Chinese Academy of Sciences, and Fang Youmin, a postdoctoral fellow, are the co-first authors of this paper, and researcher Li Hong is the corresponding author
of this paper.
Special thanks to Professor Li Xin and Engineer Ma Liangxiao of Shanghai Institute of Nutrition and Health, Chinese Academy of Sciences, and Cai Wenju, Engineer of Guizhou Science Data Center Gui'an Supercomputing Center, for their support and assistance
.
The work was supported by the National Natural Science Foundation of China, the National Key Research and Development Program of the Ministry of Science and Technology, the Youth Innovation Promotion Association of the Chinese Academy of Sciences, the Shanghai Natural Science Foundation, the Shanghai Talent Development Fund, and also supported by the biomedical big data center technology platform of the Shanghai Institute of Nutrition and Health
, Chinese Academy of Sciences.
of this paper.
Special thanks to Professor Li Xin and Engineer Ma Liangxiao of Shanghai Institute of Nutrition and Health, Chinese Academy of Sciences, and Cai Wenju, Engineer of Guizhou Science Data Center Gui'an Supercomputing Center, for their support and assistance
.
The work was supported by the National Natural Science Foundation of China, the National Key Research and Development Program of the Ministry of Science and Technology, the Youth Innovation Promotion Association of the Chinese Academy of Sciences, the Shanghai Natural Science Foundation, the Shanghai Talent Development Fund, and also supported by the biomedical big data center technology platform of the Shanghai Institute of Nutrition and Health
, Chinese Academy of Sciences.
Figure Typical framework of deep learning algorithms for drug susceptibility prediction and model evaluation process and results
Article link: https://doi.
org/10.
1093/bib/bbac605
