Cell Maps for Artificial Intelligence: AI-Ready Maps of Human Cell Architecture from Disease-Relevant Cell Lines.

This article describes the Cell Maps for Artificial Intelligence (CM4AI) project and its goals, methods, standards, current datasets, software tools , status, and future directions. CM4AI is the Functional Genomics Data Generation Project in the U.S. National Institute of Health's (NIH) Bridge2AI program. Its overarching mission is to produce ethical, AI-ready datasets of cell architecture, inferred from multimodal data collected for human cell lines, to enable transformative biomedical AI research.

Evaluation of large language models for discovery of gene set function.

Gene set analysis is a mainstay of functional genomics, but it relies on curated databases of gene functions that are incomplete. Here we evaluate five Large Language Models (LLMs) for their ability to discover the common biological functions represented by a gene set, substantiated by supporting rationale, citations and a confidence assessment. Benchmarking against canonical gene sets from the Gene Ontology, GPT-4 confidently recovered the curated name or a more general concept (73% of cases), while benchmarking against random gene sets correctly yielded zero confidence. Gemini-Pro and Mixtral-Instruct showed ability in naming but were falsely confident for random sets, whereas Llama2-70b had poor performance overall. In gene sets derived from 'omics data, GPT-4 identified novel functions not reported by classical functional enrichment (32% of cases), which independent review indicated were largely verifiable and not hallucinations. The ability to rapidly synthesize common gene functions positions LLMs as valuable 'omics assistants.

Tools for assembling the cell: Towards the era of cell structural bioinformatics.

Cells consist of large components, such as organelles, that recursively factor into smaller systems, such as condensates and protein complexes, forming a dynamic multi-scale structure of the cell. Recent technological innovations have paved the way for systematic interrogation of subcellular structures, yielding unprecedented insights into their roles and interactions. In this workshop, we discuss progress, challenges, and collaboration to marshal various computational approaches toward assembling an integrated structural map of the human cell.

Evaluation of large language models for discovery of gene set function.

Gene set analysis is a mainstay of functional genomics, but it relies on manually curated databases of gene functions that are incomplete and unaware of biological context. Here we evaluate the ability of OpenAI's GPT-4, a Large Language Model (LLM), to develop hypotheses about common gene functions from its embedded biomedical knowledge. We created a GPT-4 pipeline to label gene sets with names that summarize their consensus functions, substantiated by analysis text and citations. Benchmarking against named gene sets in the Gene Ontology, GPT-4 generated very similar names in 50% of cases, while in most remaining cases it recovered the name of a more general concept. In gene sets discovered in 'omics data, GPT-4 names were more informative than gene set enrichment, with supporting statements and citations that largely verified in human review. The ability to rapidly synthesize common gene functions positions LLMs as valuable functional genomics assistants.

Mass spectrometry profiling and quantitation of changes in circulating hormones secreted over time in Cancer borealis hemolymph due to feeding behavior.

The crustacean stomatogastric ganglion (STG) is a valuable model for understanding circuit dynamics in neuroscience as it contains a small number of neurons, all easily distinguishable and most of which contribute to two complementary feeding-related neural circuits. These circuits are modulated by numerous neuropeptides, with many gaining access to the STG as hemolymph-transported hormones. Previous work characterized neuropeptides in the hemolymph of the crab Cancer borealis but was limited by low peptide abundance in the presence of a complex biological matrix and the propensity for rapid peptide degradation. To improve their detection, a data-independent acquisition (DIA) mass spectrometry (MS) method was implemented. This approach improved the number of neuropeptides detected by approximately twofold and showed greater reproducibility between experimental and biological replicates. This method was then used to profile neuropeptides at different stages of the feeding process, including hemolymph from crabs that were unfed, or 0 min, 15 min, 1 h, and 2 h post-feeding. The results show differences both in the presence and relative abundance of neuropeptides at the various time points. Additionally, 96 putative neuropeptide sequences were identified with de novo sequencing, indicating there may be more key modulators within this system than is currently known. These results suggest that a distinct cohort of neuropeptides provides modulation to the STG at different times in the feeding process, providing groundwork for targeted follow-up electrophysiological studies to better understand the functional role of circulating hormones in the neural basis of feeding behavior.

Exploring the Sexual Dimorphism of Crustacean Neuropeptide Expression Using Callinectes sapidus as a Model Organism.

The impact of numerous diseases has been linked to differences in sex between organisms, including various neurological diseases. As neuropeptides are known to be key players in the nervous system, studying the variation of neuropeptidomic profiles between males and females in a crustacean model organism is of interest. By using high-resolution mass spectrometry with two complementary ionization sources in conjunction with quantitative chemical labeling (isotopic reductive dimethylation), differences were observed in five key neural tissues and hemolymph. Interestingly, while males and females possess numerous neuropeptide isoforms that are unique to their sex, the represented families of each sex remain largely consistent. However, some differences in familial isoforms were also observed, such as the relative numbers of neuropeptides belonging to RFamide and allatostatin A-type families. Additionally, >100 neuropeptides detected across five neural tissues and hemolymph were found to have statistically significant differences in abundance between male and female blue crab samples. Also, hundreds of putative peptide sequences were identified by de novo sequencing that may be indicative of previously undiscovered neuropeptides, highlighting the power of using a multifaceted MS approach.

Mass Spectrometry Quantification, Localization, and Discovery of Feeding-Related Neuropeptides in Cancer borealis.

The crab Cancer borealis nervous system is an important model for understanding neural circuit dynamics and modulation, but the identity of neuromodulatory substances and their influence on circuit dynamics in this system remains incomplete, particularly with respect to behavioral state-dependent modulation. Therefore, we used a multifaceted mass spectrometry (MS) method to identify neuropeptides that differentiate the unfed and fed states. Duplex stable isotope labeling revealed that the abundance of 80 of 278 identified neuropeptides was distinct in ganglia and/or neurohemal tissue from fed vs unfed animals. MS imaging revealed that an additional 7 and 11 neuropeptides exhibited altered spatial distributions in the brain and the neuroendocrine pericardial organs (POs), respectively, during these two feeding states. Furthermore, de novo sequencing yielded 69 newly identified putative neuropeptides that may influence feeding state-related neuromodulation. Two of these latter neuropeptides were determined to be upregulated in PO tissue from fed crabs, and one of these two peptides influenced heartbeat in ex vivo preparations. Overall, the results presented here identify a cohort of neuropeptides that are poised to influence feeding-related behaviors, providing valuable opportunities for future functional studies.