The leaves of 1 years old A. albus (AnBaiYu-1) ( Figure 1A ) were selected as the materials for genome analysis. The A. albus named Anbaiyu-1 ( Figures 1A–D ) was independently bred by the Ankang Academy of Agricultural Sciences. The plant was cultivated in the greenhouse of Ankang Academy of Agricultural Sciences Shaanxi province, China.

High-quality genomic DNA was extracted from A. albus leaves using an optimized cetyl-trimethylammonium bromide (CTAB) method. DNA quantity and purity were subsequently evaluated with a Qubit Fluorometer and Nano Drop spectrophotometer Genomic sequencing libraries of A. albus was performed using the DBSSEQ and PacBio HiFi platforms. For DBSSEQ sequencing, a short-read library (300-400 bp) was prepared and sequenced on the DBSSEQ platform to obtain high-quality clean reads. In parallel, a long-read library was constructed from genomic DNA and sequenced on the PacBio platform, with low-quality reads and adapter sequences removed to yield clean subreads. Genome size estimation for A. albus was conducted through K-mer distribution analysis based on DBSSEQ data. K-mer frequencies were rapidly counted using Jellyfish (v2.2.6), and then genome size, heterozygosity, repeat sequences, sequencing depth, and other genome characteristics were estimated using GenomeScope v2.01 based on the K-mer frequency results.

To assess the assembly quality of the A. albus genome, Hi-C technology was employed to facilitate contig anchoring. Assembly completeness was evaluated using Benchmarking Universal Single-Copy Orthologs (BUSCO) v10.

Genome repeat sequences were identified de novo using Repeat Modeler software (Zhao and Hao, 2007). Non-redundant LTR sequences were obtained using LTR FINDER parallel (Tarailo-Graovac and Chen, 2009). Transposable element (TE) protein sequences were predicted with Repeat Protein Mask to identify TE proteins.

Gene structure prediction was conducted through de novo, homology-based, and transcriptome-based approaches. De novo predictions were generated using Augustus software, applied to genomes masked for repeat sequences. Homology-based predictions utilized protein sequence data from Spirodela polyrhiza, Amorphophallus konjac, Selaginella moellendorffii, and Physcomitrium patens. Transcriptome-based predictions were derived from RNA-seq data, mapped to the reference genome using HISAT2 and String Tie. Genome annotation completeness was evaluated with BUSCO5 (Holt and Yandell, 2011). The gene sets from these three prediction methods were integrated using MAKER2 software (Manni et al., 2021). Functional annotation and pathway information for predicted proteins were identified by comparing sequence and motif similarities against Uni Prot, GO, KEGG, Inter Pro, Swiss Prot, Tr EMBL, Pfam, NR, GOG, KOG, and Plant TFDB databases.

Multiple approaches were employed to identify noncoding RNAs in the A. albus genome. Transfer RNAs (tRNAs) were predicted using tRNA scan-SE. Small nuclear RNA (snRNA) and ribosomal RNA (rRNA) genes were identified by searching the Rfam database with Infernal software. MicroRNA (miRNA) genes were annotated using BLASTN, referencing the miRBase dataset (Nawrocki and Eddy, 2013).

A phylogenetic tree was constructed using the genomes of 11 species: Amorphophallus albus, Amorphophallus konjac, Eucalyptus grandis, Arabidopsis thaliana, Oryza sativa, Sorghum bicolor, Populus trichocarpa, Zostera marina, Solanum tuberosum, Spirodela polyrhiza and Physcomitrium patens. Orthologous genes among A. albus and these species were identified with OrthoFinder. A maximum-likelihood (ML) phylogeny was then inferred using RAxML v8.2.12 (Stamatakis, 2014). Species divergence times were estimated with MCMC Tree v4.9, calibrated against time estimates from the Time Tree database (http://www.time-tree.org/).

Gene family expansion and contraction were analyzed using CAFE v4.2 (Han et al., 2013). Based on the species phylogenetic tree and gene family clustering results, a random birth-death model was applied to estimate the number of ancestral gene family members in each branch, predicting the contraction and expansion of gene families relative to ancestral lineages. Significant expansions or contractions were defined by p-values < 0.05, and functional enrichment analysis was conducted on gene families with significant expansion or contraction.

To investigate whole-genome duplication (WGD) events, the genomes of A. albus, A. konjac, E. grandis, A. thaliana, O. sativa, S. bicolor, P. trichocarpa, Z. marina, S. tuberosum, S. polyrhiza, and P. patens were analyzed. WGDI v0.5.6 was utilized to generate a distribution of Ks values, with Gaussian mixture models used to fit the Ks distribution curves. To assess collinearity between A. albus and A. konjac, syntenic blocks were identified using MCScanX. Genome collinearity between A. albus and related species was also evaluated with MCScanX (Wang et al., 2012). Homologous genes between species were identified through BLASTP using default parameters.

RNA-seq analyses of A. albus following SR infection were conducted at 0, 12h, 24h, and 48h post-inoculation, including control samples inoculated with PDA. Petiole samples were collected from a 1.0 cm distance from the lesion or inoculation site, with six biological replicates per treatment. All samples were immediately frozen in liquid nitrogen and stored at -80°C for later use. Sequencing was performed on the Illumina platform, with clean data obtained following quality filtering. Reads were mapped to the reference genome, and database quality was evaluated. Differentially expressed genes (DEGs) were identified across sample groups based on gene expression, annotated, and enriched. DESeq2 was used to analyze expression differences, applying a threshold of fold change (FC) ≥2 and a false discovery rate <0.1 (Love et al., 2014). DEGs were annotated and enriched using the GO and KEGG databases (Ashburner et al., 2000; Kanehisa and Goto, 2000).

Metabolite analyses of A. albus following SR infection were conducted at 0, 12h, 24h, and 48h post-inoculation, including control samples inoculated with PDA. Petiole samples were vacuum freeze-dried and analyzed using LC–MS. Raw LC–MS data files were converted to mzXML format using ProteoWizard software, and peaks with a detection rate below 50% in any sample group were discarded. Metabolite identities were assigned by referencing a custom database integrated with public databases. Variable importance in projection (VIP) values were calculated using OPLS-DA in the R package MetaboAnalystR, with score plots and permutation plots generated. Metabolite identification was performed using the KEGG compound database, and annotated metabolites were subsequently mapped to the KEGG Pathway database.

Prioritizing genes and metabolic pathways with a P-value <0.05 enables efficient data screening and facilitates rapid identification of pathways relevant to the research focus for further analysis. Genes were grouped based on differential expression, and correlation coefficients between all genes and metabolites were calculated using Pearson correlation. Data were preprocessed with Z-score transformation prior to correlation analysis, and screening was conducted based on correlation coefficient (CC) and corresponding P-values.

The genome size, repeat size and heterozygosity of A. albus were estimated using k-mer analysis. The 25-mer frequency distribution of DBSSEQ short reads showed the highest peak at a depth of 129.31. The genome size was estimated to be 5.72 Gb with 76.13% repeats. The heterozygosity rate was estimated at 2.08% ( Supplementary Table S1 ).

The genome of A. albus was sequenced using the Sequel II sequencing platform, which provided high-quality long reads for the assembly process. The sequencing data was processed to generate 3,051 contigs, which were then organized into 13 pseudo-chromosomes using Hi-C technology. Hi-C scaffolding allowed for the accurate anchoring and ordering of contigs into chromosome-scale structures. This method also helped in constructing whole-genome interaction maps, providing insights into the genome’s 3D structure and confirming the interaction patterns consistent with previously established genome interactions ( Figure 1E ). The final assembly of the A. albus genome was 5.94 Gb in size, with a high level of continuity indicated by a contig N50 of 5.61 Mb and a scaffold N50 of 467.26 Mb. These values indicate that the majority of the genome is contained within large, high-quality contigs and scaffolds. The GC content of the genome was found to be 44.87%, which is typical for plant genomes. Despite the high-quality assembly, a total of 2,246 gaps were present across the genome, with the number of gaps per chromosome varying from 62 to 321. These gaps often occur in regions of the genome that are difficult to sequence, such as those with high repetitive content or complex secondary structures ( Supplementary Table S2 ).

BUSCO analysis found 1,569 (97.21%) complete gene models and 11 (0.68%) fragmented gene models out of a total of 1,614 genes ( Supplementary Table S3 ). These results indicate that our genome assembly of A. albus is of high quality.

Using a combination of de novo and homology-based approaches, 63.83% of the assembled sequences were identified as repetitive sequences, including 58.66% LTR retrotransposons and 2.49% genome DNA transposons ( Supplementary Table S4 ).

A total of 35,155 genes were predicted using a combination of de novo, homology-based, and transcriptome-based approaches with MAKER software. The average gene length was 12,338 bp, and the average CDS length was 1,028 bp, with an average exon length of 348.38 bp ( Table 1 ; Supplementary Table S5 ).

In light of observed GCdensity, gene density, repetitive sequences density, LTR density, LINE density, DNA-TE density and Syntenic block, Two circos map of genome of A. albus was drawn ( Figures 2A, B ). Overall, the functions of 31139 genes (88.58%) were assigned. Among these, 28875(89.68%) and 16442(59.27%) had predicted homologs in NR and SwissProt databases, respectively ( Supplementary Table S6 ).

Genomes from eleven species were selected to identify homologous genes, perform gene family clustering analysis, and examine the enrichment of single-copy genes with A. albus ( Figure 2C ). The A. albus genome comprised 19,908 gene families, including 467 unique families ( Supplementary Table S7 ).

Furthermore, specific and common gene families were analyzed among A. albus and four other species. A total of 27,220 genes were clustered among A. albus, A. konjac, O. sativa, Z. marina, and S. polyrhiza, with 5,659 genes shared across all five species. A. albus and A. konjac exhibited a close relationship, sharing the highest number of gene families. Additionally, 552 specific gene families were identified in A. albus, which is fewer than the 763 specific families identified in A. konjac ( Figure 2D ).

To estimate the divergence time among the 11 species, a phylogenetic tree was constructed ( Figure 3A ). The analysis indicated that A. albus and A. konjac diverged from their common ancestor approximately 1.3 million years ago (mya).

Gene family expansion and contraction play a key role in phenotypic adaption during speciation. There were 2036 gene family contractions and 1565 gene family expansions detected in A.ablus, while 297 gene contractions and 5693 gene family expansions were identified in A.konjac ( Figure 3B ).

GO gene family enrichment analysis revealed that contracted gene families were primarily associated with ATP hydrolysis activity, heme binding, and monooxygenase activity, while expanded gene families were involved in meristem development, meristem maintenance, and DNA integration ( Supplementary Figures S1 , S2 ). KEGG gene family enrichment analysis indicated that contracted gene families were mainly related to starch and sucrose metabolism, plant-pathogen interactions, and ABC transporters, whereas expanded gene families were associated with the biosynthesis of unsaturated fatty acids, amino sugar metabolism, and phenylpropanoid biosynthesis ( Supplementary Figures S3 , S4 ).

The density distribution revealed that A. konjac shared the smallest peak with A. albus, near 0.69. Based on synonymous nucleotide substitution rates, it was determined that A. albus has undergone a whole-genome duplication (WGD) event ( Figure 4B ). Additionally, collinearity analysis of the A. albus and A. konjac genomes ( Figure 4A ) indicated that both species experienced the same WGD events.

Gene expression profiling of A. albus at different SR infection stages (0, 12h, 24h, and 48h) was conducted using RNA-Seq. Six comparative DEG analyses (S1 vs. CK, S2 vs. CK, S2 vs. S1, S3 vs. CK, S3 vs. S1, and S3 vs. S2) identified 4,576, 9,973, 5,845, 10,485, 8,677, and 1,711 differentially expressed genes (DEGs), respectively. Of these, 3,005, 3,050, 1,608, 2,645, 1,792, and 549 genes were up-regulated, while 1,571, 6,923, 4,237, 8,640, 6,885, and 1,162 genes were down-regulated in response to infection. These results clearly demonstrate that SR infection induces significant changes in plant gene expression patterns ( Figure 5A ; Supplementary Figure S5 ).

GO enrichment analysis of biological processes, cellular components, and molecular functions indicated that enriched genes were involved in the regulation of DNA-templated transcription, nucleus, and ATP binding in response to SR infection ( Figure 5B ). KEGG enrichment analysis of DEGs revealed that the response of A. albus to SR infection at the transcriptional level was primarily associated with plant hormone signal transduction, plant-pathogen interactions, and ribosome-related resistance metabolites ( Figure 5C ). GSEA analysis identified 30 biological pathways and gene sets that were differentially activated at different time points during SR infection, with 13 showing a high level of enrichment and 17 showing a lower level of enrichment ( Figure 5D ).

Metabolites in the leaves of A. albus were detected using LC-MS/MS non-targeted metabolomics at various stages of SR infection. Comparative analysis across five groups of differentially expressed metabolites (DEMs) (S1 vs. CK, S2 vs. CK, S3 vs. CK, S3 vs. S1, and S3 vs. S2) identified 151, 327, 233, 202, and 198 metabolites, respectively. Among these, 76, 143, 173, 158, and 144 metabolites were up-regulated, while 75, 84, 60, 64, and 53 metabolites were down-regulated in response to infection ( Figure 6A ). Venn diagram analysis further revealed 20 common DEMs across the five comparisons, suggesting their potential role in resistance to SR infection ( Figure 6B ). KEGG pathway analysis indicated that these DEMs were enriched in various pathways, including metabolic pathways, biosynthesis of secondary metabolites, and glycerophospholipid metabolism ( Figure 6C ). The network visually represents the interactions among differentially expressed metabolites (DEMs) in A. albus during infection, highlighting key DEMs and potential regulatory relationships that may play a significant role in response to SR infection. ( Figure 6D ).

According to the DEM analysis results of this experiment, combined with the transcriptome analysis results of the DEG, the results showed that 124 pathways were co-enriched ( Supplementary Figure S6 ). The correlations between the top 41 DEGs and 62 DEMs were selected and represented as a heat map ( Figures 7A, B ). Coenrichment annotations of DEGs and DEMs were carried out, the results of which indicated involvement in Phenylpropanoid biosynthesis, Plant hormone signal transduction, Phenylalanine metabolism, Phenylalanine and tyrosine and tryptophan biosynthesis. Meanwhile, some key metabolites (trans-4-Coumaric acid, Tyrosine, Sinapoyl aldehyde, Syringin,trans-zeatin,Tyrosine,L-Tyrosine,and Fructose 1-phosphate) in related metabolic pathways were accumulated ( Figure 8 ; Supplementary Table S8 ).