BioSample and GEO FAIR Metadata Curation with GPT-4

The repostory contains records from BioSample and GEO that relate to lung cancer, liver cancer, and ovarian cancer. he format of these files is in JSON. The raw outputs from GEO are available in the text files along with the heuristic annotations for the metadata field "tissue". They also include a curated version using carefully designed prompts that incorporate information on metadata from CEDAR using GPT-4, Medllama and Llama3.1.