Comparing reference databases for pathogen detection from inter-simple sequence repeat based sequencing of tomato seeds using Kraken2
摘要
Seedborne bacterial pathogens, including Ralstonia solanacearum (RS) and Xanthomonas euvesicatoria pv. perforans (Xep), pose significant threats to tomato production and international seed trade. Conventional detection methods such as culture-based assays and PCR are pathogen-specific and low-throughput and may fail to identify low-abundance or mixed infections. Next-generation sequencing, particularly inter-simple sequence repeat (ISSR) targeted sequencing, offers a multiplexed alternative, but accurate pathogen identification depends critically on computational classification and the choice of reference database. In this study, we evaluated six Kraken2 databases, three standard (Standard, Standard-8, and Standard-16), and three curated pathogen-focused databases (rsDB, xeDB, and mixedDB), using ISSR-derived reads from tomato seeds, including artificially inoculated samples. Standard databases were largely ineffective at reliably detecting the target pathogens, likely due to the predominance of plant-derived reads and non-target microorganisms. In contrast, curated databases enabled robust and reproducible species-level detection even at low pathogen abundance. While mixedDB allowed simultaneous detection of multiple pathogens, individual pathogen-specific databases (rsDB or xeDB) proved superior for confirmatory analyses. These findings demonstrate the potential of combining ISSR-targeted sequencing with curated Kraken2 databases for sensitive and efficient primary screening of seedborne pathogens.