BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/Chicago
X-LIC-LOCATION:America/Chicago
BEGIN:DAYLIGHT
TZOFFSETFROM:-0600
TZOFFSETTO:-0500
TZNAME:CDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0500
TZOFFSETTO:-0600
TZNAME:CST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260202T201805Z
LOCATION:241
DTSTART;TZID=America/Chicago:20251117T105000
DTEND;TZID=America/Chicago:20251117T111000
UID:submissions.supercomputing.org_SC25_sess205_ws_cafcw112@linklings.com
SUMMARY:An AI Agentic Framework for Understanding Low-Dose Radiation Effec
 ts on Human Lung Epithelial Cells
DESCRIPTION:Joshua-James Claybon (Argonne National Laboratory (ANL), Rice 
 University); Sohum Kashyap (Argonne National Laboratory (ANL), Illinois Ma
 thematics and Science Academy); Mitchell Conery, Alex Rodriguez, and Zilin
 ghan Li (Argonne National Laboratory (ANL)); John Wu (Argonne National Lab
 oratory (ANL), Harvard University); and Tarak Nandi and Ravi Madduri (Argo
 nne National Laboratory (ANL))\n\nWhile data modalities like scRNASeq, his
 tology, and DNA methylation offer valuable insights into cellular response
 s to external perturbations, learning from such datasets is often limited 
 by the user’s ability to analyze large data, and familiarity with the exis
 ting knowledge base and tools. Moreover, there is a tendency to favor well
  established mechanisms even for understanding new biology, which can limi
 t the exploration of novel or unexpected biological pathways. Manual curat
 ion of literature, pathway databases, and public datasets is time consumin
 g, and traditional analysis pipelines are typically static, tool specific,
  and lack self correcting capabilities, and are thus difficult to scale.\n
 To overcome these challenges, we present Agentic Lab, an AI agentic framew
 ork for accelerating biomedical discovery through automated, collaborative
  scientific inquiry via a set of specialized agents. Agents are entities t
 hat use prompts for understanding their tasks, LLMs for reasoning over tho
 se, and tools for interaction with the outside environment. Unlike convent
 ional linear workflows, these agents continuously reason, search, reflect,
  and adapt. For example, an unexpected gene expression pattern can automat
 ically trigger new literature searches, hypothesis refinement, or reanalys
 is of data, and coding errors and missing packages can be automatically de
 tected and fixed. Agentic Lab formulates a research workflow by first usin
 g a Principal Investigator (PI) Agent as the entry point, which interprets
  the user defined task and, with the assistance of a Browsing Agent that r
 etrieves knowledge from scientific repositories, user provided files, and 
 web links, formulates a research workflow. The PI Agent then assigns speci
 fic tasks to specialized agents. Code Writer and Executor Agents generate,
  run, and debug codes, and a Critic Agent ensures robustness through conti
 nuous evaluation of results and processes. This framework integrates liter
 ature curation, hypothesis generation, code development, and data analysis
  in iterative cycles, with the option for the user to intervene at any poi
 nt. The framework is driven entirely by open-weight LLMs that can be hoste
 d locally using limited resources, enabling local, privacy-preserving exec
 ution without reliance on costly APIs. Our approach combines smart prompti
 ng, tool augmentation, and human-in-the-loop validation to maximize the pe
 rformance of smaller models in complex biomedical discovery\nWe apply this
  framework to study low-dose (LD) radiation effects, where the carcinogeni
 c risks below 10 mGy remain poorly understood despite widespread exposure 
 from natural background (e.g., radon, cosmic rays), medical imaging, and n
 uclear industries. Using scRNA-seq data from the human lung epithelial BEA
 S-2B cell line exposed to Cs-137 gamma radiation at low (10 mGy), medium (
 100 mGy), and high (1 Gy) doses, we investigate transcriptional changes ac
 ross dose levels to identify differences in underlying biological mechanis
 ms. We use Geneformer, a pre-trained transformer-based single cell foundat
 ion model, to generate contextual gene and cell embeddings for in-silico p
 erturbation (ISP) studies and identify key drivers of cell state transitio
 ns associated with LD exposure. By analyzing shifts in the latent embeddin
 g space, we map dysregulated genes and pathways implicated in stress respo
 nse, and early malignant transformation. Agentic Lab interacts with HPC en
 vironments to submit jobs to carry out fine tuning of pretrained Genefomer
  models and ISP.\n\nRecording: Livestreamed, Recorded\n\nRegistration Cate
 gory: Technical Program Reg Pass, Workshop Reg Pass\n\nSession Chairs: Eri
 c Stahlberg (MD Anderson Cancer Center, University of Texas); Sally Elling
 son (University of Kentucky); Lynn Borkon (Frederick National Laboratory f
 or Cancer Research); Patricia Kovatch (Icahn School of Medicine at Mount S
 inai); Lauren Lewis (Frederick National Laboratory for Cancer Research); a
 nd Sean Hanlon (National Institutes of Health (NIH), National Cancer Insti
 tute (NCI))\n\n
END:VEVENT
END:VCALENDAR