Defne Circi

Large language models (LLMs) are reshaping many aspects of materials science and chemistry research, enabling advances in molecular property prediction, materials design, scientific automation, knowledge extraction, and more. Recent developments demonstrate that the latest class of models are able to integrate structured and unstructured data, assist in hypothesis generation, and streamline research workflows. To explore the frontier of LLM capabilities across the research lifecycle, we review applications of LLMs through 32 total projects developed during the second annual LLM hackathon for applications in materials science and chemistry, a global hybrid event. These projects spanned seven key research areas: (1) molecular and material property prediction, (2) molecular and material design, (3) automation and novel interfaces, (4) scientific communication and education, (5) research data management and automation, (6) hypothesis generation and evaluation, and (7) knowledge extraction and reasoning from the scientific literature. Collectively, these applications illustrate how LLMs serve as versatile predictive models, platforms for rapid prototyping of domain-specific tools, and much more. In particular, improvements in both open source and proprietary LLM performance through the addition of reasoning, additional training data, and new techniques have expanded effectiveness, particularly in low-data environments and interdisciplinary research. As LLMs continue to improve, their integration into scientific workflows presents both new opportunities and new challenges, requiring ongoing exploration, continued refinement, and further research to address reliability, interpretability, and reproducibility.

This paper investigates the use of large language models (LLMs) for extracting sample lists of polymer nanocomposites (PNCs) from full-length materials science research papers. The challenge lies in the complex nature of PNC samples, which have numerous attributes scattered throughout the text. The complexity of annotating detailed information on PNCs limits the availability of data, making conventional document-level relation extraction techniques impractical due to the challenge in creating comprehensive named entity span annotations. To address this, we introduce a new benchmark and an evaluation technique for this task and explore different prompting strategies in a zero-shot manner. We also incorporate self-consistency to improve the performance. Our findings show that even advanced LLMs struggle to extract all of the samples from an article. Finally, we analyze the errors encountered in this process, categorizing them into three main challenges, and discuss potential strategies for future research to overcome them.

Ghazal Khalighinejad, Defne Circi, L. Brinson, Bhuwan Dhingra. Findings of the Association for Computational Linguistics ACL 2024. 2024.

We report the findings of a hackathon focused on exploring the diverse applications of large language models in molecular and materials science.

Large-language models (LLMs) such as GPT-4 caught the interest of many scientists. Recent studies suggested that these models could be useful in chemistry and materials science. To explore these possibilities, we organized a hackathon. This article chronicles the projects built as part of this hackathon. Participants employed LLMs for various applications, including predicting properties of molecules and materials, designing novel interfaces for tools, extracting knowledge from unstructured data, and developing new educational applications. The diverse topics and the fact that working prototypes could be generated in less than two days highlight that LLMs will profoundly impact the future of our fields. The rich collection of ideas and projects also indicates that the applications of LLMs are not limited to materials science and chemistry but offer potential benefits to a wide range of scientific disciplines.

Understanding SARS-CoV-2 evolution is a fundamental effort in coping with the COVID-19 pandemic. The virus genomes have been broadly evolving due to the high number of infected hosts world-wide. Mutagenesis and selection are two inter-dependent mechanisms of virus diversification. However, which mechanisms contribute to the mutation profiles of SARS-CoV-2 remain under-explored. Here, we delineate the contribution of mutagenesis and selection to the genome diversity of SARS-CoV-2 isolates. We generated a comprehensive phylogenetic tree with representative genomes. Instead of counting mutations relative to the reference genome, we identified each mutation event at the nodes of the phylogenetic tree. With this approach, we obtained the mutation events that are independent of each other and generated the mutation profile of SARS-CoV-2 genomes. The results suggest that the heterogeneous mutation patterns are mainly reflections of host (i) antiviral mechanisms that are achieved through APOBEC, ADAR, and ZAP proteins, and (ii) probable adaptation against reactive oxygen species.

COVID-19 has effectively spread worldwide. As of May 2020, Turkey is among the top ten countries with the most cases. A comprehensive genomic characterization of the virus isolates in Turkey is yet to be carried out. Here, we built a phylogenetic tree with globally obtained 15,277 severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) genomes. We identified the subtypes based on the phylogenetic clustering in comparison with the previously annotated classifications. We performed a phylogenetic analysis of the first 30 SARS-CoV-2 genomes isolated and sequenced in Turkey. We suggest that the first introduction of the virus to the country is earlier than the first reported case of infection. Virus genomes isolated from Turkey are dispersed among most types in the phylogenetic tree. We find 2 of the seventeen subclusters enriched with the isolates of Turkey, which likely have spread expansively in the country. Finally, we traced virus genomes based on their phylogenetic placements. This analysis suggested multiple independent international introductions of the virus and revealed a hub for the inland transmission. We released a web application to track the global and interprovincial virus spread of the isolates from Turkey in comparison to thousands of genomes worldwide.

COVID-19 has effectively spread worldwide. As of May 2020, Turkey is among the top ten countries with the most cases. A comprehensive genomic characterization of the virus isolates in Turkey is yet to be carried out. Here, we built a phylogenetic tree with globally obtained 15,277 severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) genomes. We identified the subtypes based on the phylogenetic clustering in comparison with the previously annotated classifications. We performed a phylogenetic analysis of the first thirty SARS-CoV-2 genomes isolated and sequenced in Turkey. We suggest that the first introduction of the virus to the country is earlier than the first reported case of infection. Virus genomes isolated from Turkey are dispersed among most types in the phylogenetic tree. We find two of the seventeen sub-clusters enriched with the isolates of Turkey, which likely have spread expansively in the country. Finally, we traced virus genomes based on their phylogenetic placements. This analysis suggested multiple independent international introductions of the virus and revealed a hub for the inland transmission. We released a web application to track the global and interprovincial virus spread of the isolates from Turkey in comparison to thousands of genomes worldwide.