Pan-genome Research Notes: From Resource Construction to Functional Interpretation

1. When Do You Need a Pan-genome?

If a research question only concerns the gene structure of a single accession, local variation, or conventional SNP association analysis, a single high-quality reference genome is often sufficient. But once the focus shifts to intraspecific diversity, domestication and improvement, resistance gene repertoires, large insertions and deletions, complex NLR clusters, non-reference sequences, or population-specific genes, a single reference genome quickly becomes a limiting factor.

A pan-genome is better suited to three kinds of questions: first, how many genes and sequence variants in a species or population are not represented by the reference genome; second, whether functional module differences exist among accessions of different ecotypes, breeding stages, or geographic origins; and third, whether these differences can explain phenotype, adaptation, or disease and stress resistance.

2. Material Selection Sets the Ceiling

3. Assembly and Annotation: Lay a Solid Foundation First

Pan-genome analysis usually begins with multiple assembly versions. Before comparison, assess the completeness, contiguity, and contamination level of each genome — for example, contig/scaffold N50, BUSCO completeness, LAI, Hi-C scaffolding ratio, repeat annotation quality, and predicted gene count. If a sample is markedly fragmented, its "specific" and "absent" genes may well be technical noise rather than biology.

Annotation pipelines should also be as uniform as possible. Using different parameters, evidence sets, or software versions across samples will entangle genuine biological differences with pipeline artifacts. For crop genomes, it is recommended to integrate transcriptome evidence, protein homology, ab initio prediction, and repeat masking, and to inspect gene length, exon count, transcript count, and the proportion of functional annotations before the final comparison.

# Example quality checks before pan-genome analysis
busco -i genome.fa -l embryophyta_odb10 -m genome -o busco_genome
busco -i proteins.fa -l embryophyta_odb10 -m proteins -o busco_protein

# Keep assembly and annotation metrics in one table
# sample, assembly_size, contig_N50, scaffold_N50, BUSCO_C, gene_count, repeat_ratio

4. Choose the Right Pan-genome Representation

A pan-genome is not a single fixed format but a family of comparison frameworks. Different representations suit different questions: gene clustering is well suited to studying core and variable genes; graph pan-genomes are good at representing complex structural variation and non-linear references; reference-based PAV/SV matrices are convenient for rapid integration with population phenotypes, GWAS, and breeding materials.

5. How to Interpret Core, Variable, and Specific Genes?

Core genes are present in the vast majority of accessions and tend to be enriched in basic life processes, cellular metabolism, and development. Variable or cultivar-specific genes are more often associated with environmental adaptation, disease and stress resistance, secondary metabolism, and breeding selection. But do not equate "specific" with "important." Many specific genes may simply arise from fragmented annotation, repeat misclassification, contaminating sequences, or low-quality assembly.

A reliable interpretation should integrate multiple lines of evidence: whether the gene has a complete ORF, whether it carries a protein domain, whether there is expression evidence, whether it resides in a confidently assembled region, whether it recurs in similar materials, and whether it is consistent with phenotypic or population differentiation signals. For disease-resistance gene families such as NLRs, RLKs, and RLPs, pay particular attention to copy number variation, tandem duplication, pseudogenes, and paralog mismatches.

6. From PAV/SV to Candidate Genes

Presence–absence variation and structural variation are among the most explanatory parts of pan-genome research. Compared with SNPs, PAVs and SVs are more likely to directly affect gene dosage, promoter structure, coding-sequence integrity, and local chromosome architecture. For crop disease resistance, quality, plant architecture, and environmental adaptation traits, many key differences are not single-nucleotide mutations but large insertions, deletions, inversions, transposon insertions, or gene cluster rearrangements.

1. First confirm that the variant is real, avoiding false positives caused by assembly breakpoints, repeat regions, or alignment errors.
2. Project the variant onto genes, promoters, repeats, and conserved domains to assess its potential functional impact.
3. Build a PAV/SV matrix and jointly analyze it with phenotype, population structure, geographic origin, and domestication grouping.
4. Within candidate regions, integrate expression, haplotype, selective sweep, known QTL, and homologous gene evidence.
5. Validate key variants using independent materials, PCR markers, long-read validation, or functional experiments.

7. Pan-genome, GWAS, Haplotypes, and Breeding Applications

A pan-genome should not stop at the level of resource description. A more valuable approach is to convert pan-genome variation into a marker system that is analyzable, verifiable, and usable. Beyond SNP-GWAS, one can also carry out PAV-GWAS, SV-GWAS, gene copy number association analysis, gene presence frequency comparisons, superior haplotype mining, and regional combination pattern analysis.

In breeding applications, the point is not to list every variant but to filter for variants that are stable, detectable, of clear effect, and low in background dependency. For disease resistance traits, focus on genes that are present at high frequency in resistant materials and absent or functionally disrupted in susceptible materials; for complex quantitative traits, it is more appropriate to integrate haplotype effects, expression levels, and phenotypic stability across multiple environments for a combined ranking.

8. A Recommended Analysis Workflow

Closing Remarks

Pan-genome research is most prone to two pitfalls: the first is chasing sample counts and map size while neglecting data quality; the second is stopping at statistics on the numbers of core and variable genes without interpretation oriented toward traits. A truly valuable pan-genome study should design its materials and technical route around a clear question from the very beginning, and leave interfaces for downstream functional validation and breeding use.

For crop researchers, the ultimate goal of a pan-genome is not a more complicated genome map, but a more accurate understanding of how genetic diversity shapes phenotypic differences. Only by connecting assembly, annotation, structural variation, population genetics, phenotypic data, and experimental validation can a pan-genome progress from resource construction to biological discovery.