Background

Single-nucleotide polymorphism (SNP) discovery and genotyping are fundamental to breeding research [1,2]. However, whole-genome sequencing for this purpose is often too costly and yields a high number of redundant SNPs. In response, reduced-representation sequencing (RRS) has emerged as a more efficient alternative. This strategy involves selectively sequencing only a small, targeted fraction (typically 1%–10%) of the genome, enabling efficient genome-wide marker detection for breeding programs [3,4].

As a classical RRS method, restriction site-associated DNA sequencing (RAD-seq) follows a multi-step protocol. The workflow starts with restriction enzyme (RE) digestion of genomic DNA and ligation of a barcoded P1 adapter. After pooling individually barcoded samples, the DNA is randomly sheared and size-selected. Finally, a P2 adapter is ligated, followed by PCR amplification and sequencing [5]. While this approach effectively captures partial genomic regions and has been widely applied in genetic and breeding studies across diverse species, its multi-step manual procedures—including fragment shearing, purification, and adapter ligation—render the library preparation process cumbersome, which has spurred ongoing optimization efforts [5,6]. Subsequent derivative methods, such as 2b-RAD, have been developed to simplify the protocol. They utilize type-IIB restriction enzymes, which cleave DNA both upstream and downstream of their recognition sites, producing uniform fragments of approximately 30 bp. This inherent uniformity removes the need for separate shearing and size selection, thereby streamlining library preparation [7,8]. Double-digest RAD (dd-RAD) has introduced improvements by eliminating random shearing and implementing defined size selection, thereby restricting sequencing to regions adjacent to RE recognition sites [9]. Furthermore, ezRAD [10] and MethylRAD [11] have also been developed to further streamline workflows or target specific genomic features. Nevertheless, these all require restriction enzyme digestion first, followed by sequencing library construction through adapter ligation or TA ligation. These methods still involve labor-intensive, multi-step separation and purification, requiring significant time and resources—especially for large sample sets. Furthermore, the single-sample size selection used in ezRAD and ddRAD can cause inter-sample variability in fragment recovery. This variability arises from the randomness of the selection process and can lead to marker loss and reduced reproducibility [6].

We leverage the simplicity and high efficiency of the Tn5 transposase, which performs simultaneous DNA fragmentation and adapter ligation [12,13]. Building on this foundation, we integrated it with our previously developed all-in-one sequencing (AIO-seq) method. AIO-seq enables pooled PCR amplification, size selection, and quantification for hundreds of samples [14]. On this basis, we introduce an innovative RRS strategy termed inverse restriction site–associated DNA sequencing (iRAD-seq). This approach fundamentally reverses the conventional workflow by adopting an inverse sequence: constructing the sequencing library first and subsequently selecting target fragments (Figure 1). iRAD-seq integrates Tn5 transposase to achieve one-step DNA fragmentation and adapter tagging. This innovation dramatically simplifies library construction, boosts operational efficiency, and ultimately delivers a more streamlined, high-throughput, and user-friendly solution for RRS library preparation.