Related Experiment Video
Updated: Sep 1, 2025
09:20
Isolation and Differentiation of Adipose-Derived Stem Cells from Porcine Subcutaneous Adipose Tissues
Published on: March 31, 2016
29.4K
Using Machine Learning to Identify Biomarkers Affecting Fat Deposition in Pigs by Integrating Multisource
Huatao Liu1, Kai Xing1, Yifan Jiang1
1National Engineering Laboratory for Animal Breeding, Laboratory of Animal Genetics, Breeding and Reproduction, Ministry of Agriculture, College of Animal Science and Technology, China Agricultural University, Beijing 100193, China.
Journal of Agricultural and Food Chemistry
|August 11, 2022
Summary
Machine learning, specifically AdaBoost, accurately predicted pig fat content using transcriptome data. This approach identified 12 key genes, including FASN and APOD, crucial for understanding fat deposition in pigs and human obesity models.
Area of Science:
- Animal Science
- Genomics
- Bioinformatics
Background:
- Fat deposition in pigs impacts production efficiency and pork quality, serving as a model for human obesity.
- Transcriptome sequencing is valuable for studying fat deposition, but traditional methods face challenges with small sample sizes and high false positive rates.
- Novel analytical strategies are needed to improve the accuracy and consistency of transcriptome data analysis for fat deposition studies.
Purpose of the Study:
- To develop and validate a machine learning approach for analyzing pig transcriptome data to identify genes associated with fat deposition.
- To compare the performance of eight different machine learning methods in predicting high and low fat content groups.
- To identify potential candidate genes that influence fat deposition in pigs.
Main Methods:
- Collected and sequenced mRNA from 36 adipose, muscle, and liver tissue samples from Songliao black and Landrace pigs.
- Integrated 64 additional transcriptome datasets from the GEO database, creating a combined dataset of 100 samples.
- Applied machine learning algorithms, including AdaBoost, to analyze standardized transcriptome data, utilizing differential gene expression features and cross-validation.
Main Results:
- AdaBoost demonstrated superior performance among eight machine learning methods, achieving over 93% prediction accuracy and the highest area under the curve in predicting fat content groups.
- The AdaBoost model identified 12 genes associated with fat deposition based on their performance ranks.
- Specific genes like FASN and APOD showed adipose tissue-specific expression, while APOA1 was liver-specific, suggesting their roles in fat metabolism.
Conclusions:
- Machine learning, particularly AdaBoost, offers a robust and accurate strategy for analyzing complex transcriptome data to understand fat deposition.
- The identified genes, including FASN, APOD, and APOA1, represent promising candidate biomarkers for regulating fat deposition in pigs.
- This study provides valuable insights for improving pig breeding strategies and offers a potential model for human obesity research.
