The Fly Base Annotation Helper is a Python script that helps annotate Drosophila genes in scientific papers. Given a list of PubMed IDs (pmids), the script downloads the corresponding papers, parses their content, and identifies genes that appear in italics, in non-italicized text, and in the main sections of the paper (excluding the introduction).
The script outputs a TSV file with each line corresponding to a gene occurrence in the paper, unless the configuration states that no gene occurrence or snippet should be output, in which case each line corresponds to one gene in one paper. The output includes the following columns:
PMID of the paper FBGNID of the gene Gene occurrence (the actual spelling of the gene in the text), if desired Snippet (can be configured for long, short, or none) Gene frequency (number of gene occurrences in the paper that are the given gene, if desired) Word frequency (number of words in the paper that are the gene, if desired) Raw count (number of occurrences of the gene in the paper, if desired) The last three columns can be used as a confidence measure.
Before running the Fly Base Annotation Helper, make sure that you have Python 3 installed on your system. You can download it from the official website. You will also need to install the required dependencies listed in the requirements.txt file by running the following command in your terminal:
pip install -r requirements.txt
To use the Fly Base Annotation Helper, you first need to generate the necessary resources by running the update_resources.py script. This script looks up the necessary FlyBase files from config.ini and generates two pickles: a pmid to pmcid dictionary and a gene dictionary that maps different spellings of the genes to their FBGNID.
Once the resources are generated, you can run the main script fly_base_annotation_helper.py. The script can be configured via config.ini. You will also need to pass it a file containing one pmid per line, for the papers for which you want the genes to be found. The configuration file includes options to specify paths to FlyBase files, paths to pickles, output paths, and various parameters, such as the politeness parameter for requests to the NCBI server and whether to output gene occurrence, snippet, gene frequency, word frequency, and raw count. If you use the machine learning algorithm, the output will be paper, gene and confidence.
The deep learning model can be found at hugging face FlyBaseGeneAbstractClassifier
The output of the Fly Base Annotation Helper is a TSV file with the columns described above. The output can be used help human annotators tag the papers with the genes that they are about.
If you have any questions or issues with the Fly Base Annotation Helper, please feel free to open an issue on the GitHub repository. Contributions are also welcome via pull requests.
The Fly Base Annotation Helper is released under the GPL 3.0 License.