cgrivaz/FlyBaseGeneAbstractClassifier

1---2license: mit3widget:4- text: "'nord 174 nord. Hedgehog (Hh) and Bone Morphogenetic Proteins (BMPs) pattern the developing Drosophila wing by functioning as short- and long-range morphogens, respectively. Here, we show that a previously unknown Hh-dependent mechanism fine-tunes the activity of BMPs. Through genome-wide expression profiling of the Drosophila wing imaginal discs, we identify nord as a novel target gene of the Hh signaling pathway. Nord is related to the vertebrate Neuron-Derived Neurotrophic Factor (NDNF) involved in congenital hypogonadotropic hypogonadism and several types of cancer. Loss- and gain-of-function analyses implicate Nord in the regulation of wing growth and proper crossvein patterning. At the molecular level, we present biochemical evidence that Nord is a secreted BMP-binding protein and localizes to the extracellular matrix. Nord binds to Decapentaplegic (Dpp) or the heterodimer Dpp-Glass-bottom boat (Gbb) to modulate their release and activity. Furthermore, we demonstrate that Nord is a dosage-dependent BMP modulator, where low levels of Nord promote and high levels inhibit BMP signaling. Taken together, we propose that Hh-induced Nord expression fine-tunes both the range and strength of BMP signaling in the developing Drosophila wing.'"5---6# FlyBaseGeneAbstractClassifier7 8This repository hosts the `FlyBaseGeneAbstractClassifier`, a machine learning model designed to classify gene-paper abstract pairs into two labels for Drosophila genes: 9 10- LABEL_1: The gene is a topic of the paper.11- LABEL_0: The gene is not a topic of the paper.12 13The model was trained on a dataset made from open papers tagged by FlyBase as of February 2022. The training data set consists of 43,000 gene-abstract pairs, and was tested on 4,846 gene-abstract pairs.14 15## Requirements16 17The model requires the `transformers` library and was trained on a system with the following hardware:18 19- CPU count: 620- GPU count: 121- GPU type: NVIDIA A100-SXM4-40GB22 23## Usage24 25To use the model, follow these steps:26 27```python28from transformers import AutoTokenizer, AutoModelForSequenceClassification, pipeline29 30# Initialize the tokenizer and the model31tokenizer = AutoTokenizer.from_pretrained("scibert", model_max_length=512)32model = AutoModelForSequenceClassification.from_pretrained("cgrivaz/FlyBaseGeneAbstractClassifier", num_labels=2)33 34# Create a pipeline35model_pipeline = pipeline("sentiment-analysis", model=model, tokenizer=tokenizer)36```37 38## Training and Evaluation39 40Detailed information about the training process and evaluation metrics can be found on the project's Weights & Biases page [here](https://wandb.ai/cgrivaz/gene_tagging/runs/38574bvw).41 42## Limitations and Future Work43 44As the model is in its initial version, it is likely that there are areas for improvement and potential biases that have not been thoroughly investigated. Users are encouraged to provide feedback and report any issues they encounter during usage.45 46## Contributing47 48Contributions to improve the model are welcome. Please feel free to open an issue or submit a pull request.49 50## License51 52mit53 54