{"categories":[{"label":"Artificial Intelligence","url":"https://skillfed.io/packages/category/scientific-engineering-artificial-intelligence/9"}],"enrichment":{"capability":"AntiBERTy is a transformer language model pre-trained on antibody sequences that generates embeddings, classifies antibody species and chain types, predicts masked residues, and scores sequence likelihood.","skillfed_tags":["antibody-biology","transformer-model","protein-embedding"],"use_cases":["Generate fixed-size vector representations of antibody sequences for clustering, similarity search, or downstream ML models.","Classify antibody sequences by species origin and chain type automatically.","Predict missing or masked residues in antibody sequences based on learned patterns.","Score the likelihood of antibody sequences to identify natural vs. synthetic variants.","Analyze attention patterns across antibody sequences to understand model reasoning."],"what_it_does":"AntiBERTy is a BERT-style transformer model trained on natural antibody sequences. It provides a pre-trained encoder for antibody protein analysis, allowing you to extract learned representations of antibody sequences without training from scratch. The package wraps the model in a simple Python interface with methods to generate fixed-dimension embeddings, classify antibody species and chain type, fill in masked positions, and score sequence likelihood.\n\nYou instantiate an AntiBERTyRunner, then call methods like embed(), classify(), fill_masks(), or pseudo_log_likelihood() on lists of antibody sequences. It depends on torch and transformers for the underlying model and inference. The model is designed for computational biology workflows where you need to work with antibody sequences programmatically\u2014for example, screening variants, analyzing affinity maturation, or featurizing sequences for downstream machine learning.","worth_installing":"Yes, if you work with antibody sequences and need a pre-trained transformer encoder\u2014the model is specialized and the install is frictionless. No, if you need active maintenance or support; the package is abandoned and has no license metadata. Consider it a research artifact: useful for one-off analysis or as a feature extractor in a larger pipeline, but not for production systems requiring updates or legal clarity."},"id":"antiberty","links":{"html":"https://skillfed.io/packages/antiberty","md":"https://skillfed.io/packages/antiberty.md","pypi":"https://pypi.org/project/antiberty/"},"maintenance":{"status":"abandoned"},"meta":{"latest_release":"2023-07-16","license_spdx":null,"license_treatment":"unclear","name":"antiberty","python_support":"unspecified","summary":null},"popularity":{"monthly_downloads":120393,"position":12025,"tier":"top_15000"},"security":{"n_vulnerabilities":0},"version":"0.1.3"}