DEA for Bioinformaticians

What it does

The app is useful to find the most representative gene given a sample. Differential expression analysis is a method to decipher the interaction between genes and their environment.

The application is made under a service-based architecture. The “front” uses the Flutter SDK, coded in “Dart”, the graphical interface requests the CSV file where the samples are and you want to find the most representative gene on that “dataset”.

The “Backend” is coded in Python using “FastAPI” libraries and the “Uvicorn” web server. The magic of Gemini is in the back, using the “Langchain” framework. In a first phase, the most representative genes of the different “chunks” are found. Once the list of the most representative genes of the different “chunks” is available, Gemini is asked to find the most representative gene, giving a description of the process that was carried out and an explanation of the gene found.