Motivation: In cellular biology, researchers design wet experiments by reading the relevant articles and considering the described experiments and results. Today, researchers spend a long time exploring the literature in order to plan experiments. Results: To accelerate experiment planning, we have developed a web application named LEXAS (Life-science EXperiment seArch and Suggestion). LEXAS curates the description of biomedical experiments and suggests the experiments on genes that could be

more »

... formed next. To develop LEXAS, we first retrieved the descriptions of experiments from full-text biomedical articles archived in PubMed Central. Using these retrieved experiments and biomedical knowledgebases and databases, we trained a machine learning model that suggests the next experiments. This model can suggest not only reasonable genes but also novel genes as targets for the next experiment as long as they share some critical features with the gene of interest. Availability and implementation: LEXAS is available at https://lexas.f.u-tokyo.ac.jp/ and provides users with two interfaces: search and suggestion. The search interface allows users to find a comprehensive list of experiment descriptions, and the suggestion interface allows users to find a list of genes that could be analyzed along with possible experiment methods. The source code is available at https://github.com/lexas-f-utokyo/lexas.