Multilingual translation for zero-shot biomedical classification using BioTranslator

Abstract

Existing annotation paradigms rely on controlled vocabularies, where each data instance is classified into one term from a predefined set of controlled vocabularies. This paradigm restricts the analysis to concepts that are known and well-characterized. Here, we present the novel multilingual translation method BioTranslator to address this problem. BioTranslator takes a user-written textual description of a new concept and then translates this description to a non-text biological data instance. The key idea of BioTranslator is to develop a multilingual translation framework, where multiple modalities of biological data are all translated to text. We demonstrate how BioTranslator enables the identification of novel cell types using only a textual description and how BioTranslator can be further generalized to protein function prediction and drug target identification. Our tool frees scientists from limiting their analyses within predefined controlled vocabularies, enabling them to interact with biological data using free text.

Evaluating the Utilities of Foundation Models in Single-cell Data Analysis

“…While LLMs have seen marked success in the realms of DNA analysis [25] and biomedical NLP [26, 27], their application in single-cell research remains largely uncharted. There is a limited number of robust pre-trained models (known as single-cell LLMs) capable of managing multiple tasks in single-cell research.…”

James German and the Quest to Understand Human RECQ Helicase Deficiencies

“…The ability to integrate and explore these different types of data across different temporal and spatial scales will be powerfully amplified by AI and machine learning-enabled approaches [124][125][126]. Low-hanging fruit is already in sight with the identification of new associations among incompletely mapped or annotated data starting from simple text-based queries [127], where even simple gene expression profiling data (e.g., Figure 6) can provide useful and revealing points of departure.…”

BioT5: Enriching Cross-modal Integration in Biology with Chemical Knowledge and Natural Language Associations

“…Cross Text-protein Modalities ProteinDT (Liu et al, 2023a) is a multi-modal framework that uses semantically-related text for protein design. BioTranslator (Xu et al, 2023a) is a cross-modal translation system specifically designed for annotating biological instances, such as gene expression vectors, protein networks, and protein sequences, based on user-written text. Cross Three or More Biology Modalities Galactica (Taylor et al, 2022) is a general GPT-based large language model trained on various scientific domains, including scientific paper corpus, knowledge bases (e.g., PubChem (Kim et al, 2023) molecules, UniProt (uni, 2023) protein), codes, and other sources.…”