Machine learning model identifies cancer-specific enhancer-gene interactions
2023
Weill Cornell Medicine, New York, USA
In this study, a computer model called Differential Gene Targets of accessible chromatin (DGTAC) was developed. It helps identify specific genetic areas that promote the growth of malignant cells, which is important for the development of anti-cancer drugs. The model uses data from 371 patients with different types of cancer to identify these areas. It only needs a small amount of tissue from the patients to make reliable predictions.
The model uses specific data (ATAC-seq and RNA-seq) to predict how genes are activated. It also takes into account other information such as the distance of the genetic sequences from their transcription start sites and how strongly they are activated. An important point in the predictions is a special error value that is calculated for each sample.
In tests with different cancer cells, the model was able to identify new areas that influence the activity of 602 cancer genes. In addition, the predictions of the model were checked and confirmed in experiments. The model can also distinguish different types of genetic regions, which previous methods could not.
The results of this study help us to better understand the genetic causes of diseases triggered by faulty gene activity. In addition, the model shows how it can be useful in developing tailored drugs for individual patients through its patient-specific approach.
Recapitulation of patient-specific 3D chromatin conformation using machine learning
Ekta Khurana
Added on: 09-26-2023
[1] https://www.sciencedirect.com/science/article/pii/S2667237523002229?via%3Dihub[2] https://www.cell.com/cell-reports-methods/fulltext/S2667-2375(23)00246-1?dgcid=raven_jbs_etoc_email