Fatemeh Zare-Mirakabad - Computational Biology Research Center (CBRC), Department of Mathematics and Computer Science, Amirkabir University of Technology, Tehran, Iran
Zahra Seraj - Computational Biology Research Center (CBRC), Department of Mathematics and Computer Science, Amirkabir University of Technology, Tehran, Iran
Bahareh Attaran - Department of Microbiology, Faculty of Biological Sciences, Alzahra University, Tehran, Iran

Acinetobacter baumannii is a type of gram-negative bacillus frequently found in hospital settings and among hospitalized patients. This bacterium has the ability to infiltrate open wounds. Its shape resembles a combination of a rod and a ball. It has rapidly developed resistance to numerous antibiotics.Antimicrobial resistance (AMR) stands as a significant global challenge that poses a threat to human and animal health. The urgent necessity for rapid and precise AMR diagnostic techniques is evident. Traditional methods, such as antimicrobial susceptibility testing (AST), are time-consuming, have limited throughput, and can only be applied to cultivable bacteria.However, machine learning (ML) approaches present promising opportunities to tackle this issue by enabling automated prediction of AMR using bacterial genomic data. In most ML approaches, the input consists of the bacterial genome and information regarding the antibiotic under consideration. The ML models then predict whether the bacteria are resistant to the specific antibiotic.Still a significant challenge in this field is the limited availability of bacterial data, which results in difficulties in training robust and accurate ML models. Addressing this data scarcity is crucial to enhance the performance and reliability of AMR prediction models.To address the challenge of limited data, this study introduces a novel approach utilizing a Siamese neural network. The network is designed to take two bacterial genomes as inputs, aiming to predict whether these bacteria demonstrate resistance to a specific antibiotic. The Siamese architecture enables a comparative analysis between the genomes, allowing the model to capture nuanced variations and patterns that potentially contribute to antimicrobial resistance.By leveraging this innovative network design, our goal is to significantly improve the accuracy and efficiency of predicting bacterial resistance profiles. Ultimately, this approach can aid in the identification of effective antibiotic treatments, thereby mitigating antimicrobial resistance