Banca de DEFESA: PEDRO VICTOR BARBOSA ARAUJO

Uma banca de DEFESA de MESTRADO foi cadastrada pelo programa.
STUDENT : PEDRO VICTOR BARBOSA ARAUJO
DATE: 26/08/2026
TIME: 14:00
LOCAL: Google Meet.
TITLE:

FROM DATA CURATION TO RISK REPORTING: A PIPELINE FOR POLYGENIC RISK SCORES 

KEY WORDS:

Polygenic risk scores; PRS; PRS report; phenotypic traits; genetic variants; precision medicine; GWAS; PLINK2; language models; clinical recommendations

 

PAGES: 184
BIG AREA: Ciências Biológicas
AREA: Biologia Geral
SUMMARY:

Polygenic Risk Scores (PRS) have emerged as a promising approach for assessing the
genetic risk of complex diseases by aggregating the effects of multiple genetic variants
associated with a given phenotype. Despite their potential, the practical application of
PRS still faces challenges related to methodological standardization, clinical interpretation
of results, and the integration of these data into reports that are accessible to healthcare
professionals.

In this work, an automated computational pipeline is proposed and implemented for the
generation of PRS reports, encompassing data acquisition and genomic data processing
through to the structured presentation of results and the generation of personalized clinical
recommendations for multiple phenotypic traits, enabling the simultaneous analysis of
different conditions and complex characteristics within the same individual. The pipeline
relies on publicly available data from genome-wide association studies (GWAS) and the
1000 Genomes Project as a reference population, employing PLINK2 for score calculation,
result normalization, and the derivation of comparative metrics such as z-scores and
percentiles.

Beyond PRS calculation, the system performs phenotype categorization, automatic generation of graphical visualizations, and individual risk interpretation across different risk
levels. The clinical recommendation generation stage uses large language models (LLMs)
to interpret each phenotype based on its description and estimated risk level. Recommendations are pre-structured and stored in JSON format, ensuring reusability, consistency, and seamless integration into the final report.

The results demonstrate that the proposed pipeline enables the production of reproducible,
interpretable, and integrable PRS reports, contributing to the systematic use of PRS as a
decision-support tool in precision medicine while acknowledging its methodological and
clinical limitations.

 

COMMITTEE MEMBERS:
Presidente - 1939184 - SANDRO JOSE DE SOUZA
Interno - 1267860 - GUSTAVO ANTONIO DE SOUZA
Externo à Instituição - GILDERLANIO SANTANA DE ARAÚJO