Abstract 2506: Microhunt: an integrated pipeline for precise microbial quantification in low-biomass clinical samples

Abstract Accurately quantifying microbial abundance from high-throughput sequencing data is challenging, especially in low-biomass samples where host DNA/RNA dominates microbial content. Host or environmental contaminants may obscure true microbial signals, leading to false positives. Here, we introduce MicroHunt, an advanced pipeline integrating quality control, host sequence depletion, microbial classification, and contaminant exclusion.MicroHunt incorporates comprehensive human reference genomes (hg38, T2T-CHM13, Gencode) to efficiently remove host-derived reads and eliminates contaminant sequences source, such as vectors, phages, and plasmids. It supports multiple microbial classification tools and utilizes a curated reference database encompassing archaea, fungi, bacteria, viruses, and protozoa. By incorporating host genome references into the classification process, MicroHunt clearly distinguishes host from microbial sequences, significantly improving classification accuracy and reducing false positives. MicroHunt also provides integrated contaminant control strategies effective even without negative controls, enabling reliable microbial quantification from low-biomass samples. Using ribosomal RNA-depleted whole-transcriptome sequencing data from 807 colorectal cancer (CRC) tumor samples in the AVANT Phase III trial, MicroHunt effectively characterized the dominant microbial phyla present within the tumor microbiome of CRC patients, with Pseudomonadota and Bacillota being the most prevalent. Further analysis revealed that CRC patients with high abundance of Lactobacillus iners (L. iners) in tumor tissues were significantly associated with poor prognosis following chemotherapy combined with bevacizumab treatment (Hazard Ratio = 2.21, 95% CI = 1.14-4.32, p = 0.017). Sequence alignment analysis confirmed that the reads identified as L. iners were not derived from potential contamination sources. Moreover, integrating microbiome data generated by MicroHunt with patient transcriptomic data uncovered complex interactions between the gut microbiome and host gene expression in CRC patients. Pathway enrichment analysis revealed a significant association between L. iners and the fatty acid oxidation pathway (Spearman’s correlation coefficient r = 0.094, p = 0.0079), suggesting that microbial metabolism may influence patient prognosis in CRC. Finally, we validated the utility of MicroHunt using sequencing data from other tumor types, including non-small cell lung cancer and bladder cancer, successfully identifying intratumoral microbes and elucidating their potential biological significance. Citation Format: Gengwen Deng, Lingxuan Zhu, Weiming Mou, Liling Xu, Shujia Lin, Keren Wu, Hao Li, Zhixing Deng, Hongsen Huang, Zeji Guo, Qiheng Liang, Aimin Jiang, TIng Wei, Jiang Zhang, Quan Cheng, Anqi Lin, Peng Luo. Microhunt: an integrated pipeline for precise microbial quantification in low-biomass clinical samples [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 2506.