https://github.com/ggonnella/egcgpt
Extract rules about the contents of prokaryotic genomes from scientific literature using GPT
https://github.com/ggonnella/egcgpt
Last synced: 3 months ago
JSON representation
Extract rules about the contents of prokaryotic genomes from scientific literature using GPT
- Host: GitHub
- URL: https://github.com/ggonnella/egcgpt
- Owner: ggonnella
- Created: 2023-05-10T14:11:27.000Z (about 2 years ago)
- Default Branch: main
- Last Pushed: 2023-05-15T20:29:18.000Z (about 2 years ago)
- Last Synced: 2025-01-16T16:23:20.172Z (4 months ago)
- Language: Python
- Size: 90.8 KB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# EGC-GPT
This directory contains prompts for attempting automatic extraction, using GPT
large language models, of rules about the contents of genomes of prokaryotic
organisms from scientific literature.
## Modules
The modules directory contains collection of modules, i.e. different
way to formulate the same concepts (e.g. more or less verbose).
## Prompt configuration files
The prompts directory contains collection of prompt configuration files,
YAML files, which express a prompt as combination of entries from module files.
Paths of the module files are written relative to the main directory of the
repository.
## Script for combining text modules
The scripts directory contains the script ``combine_text_modules.py``,
which is employed for combining small snippets of text
(here called modules) into a complete text.
The script takes as an argument the prompt configuration file and outputs
a prompt, which can be passed to GPT.
Since paths in the prompt configuration files are written as relative to
the main directory of the repository, the way to create a prompt is:
```
cd egcgpt # main directory of the repository
scripts/combine_text_modules.py prompts/FILENAME
```
where FILENAME is the name of the prompt configuration file.
## Acknowledgements
This rule collection has been created in context of the DFG project GO 3192/1-1
“Automated characterization of microbial genomes and metagenomes by collection
and verification of association rules”. The funders had no role in study
design, data collection and analysis.