Page 148 - SRPSKO DRUŠTVO ISTRAŽIVAČA RAKA

Serbian Association for Cancer Research SDIRSACR

P55

In Silico Characterization of the Non-Coding Transcript BUD23-212 in Gastrointestinal Cancers

Nikola Krizmanic, Aleksandra Nikolic, Tamara Babic

Institute of Molecular Genetics and Genetic Engineering, University of Belgrade, Belgrade, Serbia

Keywords: Gastrointestinal Neoplasm; Gene Expression Regulation, Neoplastic; MicroRNAs; RNA, Long Noncoding;
Transcriptome

Background: Recent pan-cancer transcriptomic analyses have identified differential activity of two alternative promoters
of the BUD23 gene in malignant versus non-malignant gastrointestinal (GIT) mucosa. The promoter upregulated in
tumor tissues drives the expression of the transcript isoform BUD23-212 (ENST00000453316), suggesting a potential
role in malignant transformation. This study aimed to predict the functional relevance of BUD23-212 in gastrointestinal
cancers using an in silico approach.
Methods: We employed publicly available in silico tools to evaluate the transcript’s coding potential, subcellular
localization, repetitive element content, and miRNA binding interactions. Transcript expression profiles in tumor and
non-tumor samples from the esophagus, stomach, colon, and rectum were retrieved from the UCSC Xena browser.
Additionally, we assessed BUD23-212 expression in non-malignant HCEC1CT cells and malignant cell lines HCT116,
DLD1, SW620, and DLD1R using our transcriptomic dataset.
Results: CPC2 classified BUD23-212 as non-coding with high probability. According to AnnoLnc2, the transcript contains
Alu/SINE repetitive elements, which are often enriched in non-coding RNAs, and confer regulatory and structural
functions. lncLocator predicted BUD23-212 enrichment in the nucleus and ribosomes, suggesting potential roles in
epigenetic regulation, RNA processing, or micropeptide translation. miRBase and miRDB predicted binding sites for
miR-1285-2, miR-509 isoforms, and miR-4308, microRNAs with established tumor-suppressive roles in gastric and
colorectal cancers, suggesting that BUD23-212 may function as a competitive endogenous RNA (ceRNA) or molecular
sponge. According to UCSC Xena data, BUD23-212 expression was significantly elevated in the tumor compared to non-
tumor GIT tissues. Our transcriptomic data confirmed expression in malignant cell lines HCT116 and DLD1R.
Conclusions: BUD23-212, which is overexpressed in malignant GIT tissues and analyzed cancer cell lines, appears to
exert its molecular function as a regulatory non-coding RNA. Future research should aim to clarify the precise molecular
functions of BUD23-212, explore its potential as a biomarker, and investigate the therapeutic potential of its targeted
silencing in gastrointestinal cancers.

Acknowledgments and funding: This research was supported by the Science Fund of the Republic of Serbia, PROMIS,
#6052315, SENSOGENE and IMGGE Annual Research Program for 2025, Ministry of Science, Technological Development
and Innovation of the Republic of Serbia, 451-03-136/2025-03/200042.

P56
Comparative analysis of machine learning algorithm results in predicting the probability of rectal tumor
presence

Aleksandra Bibić¹, Ivana Mišković², Stevan Pecić³, Zorica Nestorović¹, Mladen Marinković ², Edib Dobardžić³
1,
1Faculty of Medicine, University of Belgrade, Belgrade, Serbia
2 Clinic for Radiation Oncology, Institute for Oncology and Radiology, Belgrade, Serbia
3 Faculty of Physics, University of Belgrade, Belgrade, Serbia

Keywords: algorithms, computed tomography, machine learning, rectal carcinoma

Background: This study examines the potential applications of machine learning (ML) algorithms in the analysis of
computed tomography (CT) scans aimed at diagnosing tumor changes in specific organs. The focus of this study is on
detecting tumor tissue in the rectal area. The aim is to train a model that will recognize and identify the tumor location
using exclusively CT scans, without relying on magnetic resonance imaging. The long-term goal is the development
and validation of algorithms that could be integrated into medical devices for automatic identification of tumor tissue,

133