BMC Genomics最新文献

{"title":"A large palindromic segmental duplication on mouse chromosome 12 marks an evolutionarily fragile genomic interval.","authors":"Yong-Kook Kang","doi":"10.1186/s12864-026-13210-y","DOIUrl":"10.1186/s12864-026-13210-y","url":null,"abstract":"<p><strong>Background: </strong>Palindromic segmental duplications (SDs), in which duplicated sequence arms flank a central spacer in inverted orientation, are among the most structurally consequential configurations in mammalian genomes, yet their occurrence in the mouse genome remains poorly characterized. Here, SD12A1 is systematically characterized as a large palindromic SD locus on mouse chromosome 12 (chr12:17.95-24.51 Mb, mm10).</p><p><strong>Results: </strong>Analysis of 9,263 UCSC mm10 pairwise SD entries, integrated with de novo CpG island (CGI) profiling, directional sequence matching, block-level sequence identity analysis, and mouse-rat-human synteny comparison, showed that SD12A1 comprises two SD-rich arms flanking a 530 kb SD-depleted central gap. Inter-arm SD pairs were predominantly inverted, whereas intra-arm pairs were predominantly direct. Reverse-complement CGI analysis identified 18 left-arm CGIs with high-confidence right-arm matches (best-match RC cosine similarity > 0.75), and directional matching of 50 bp Block 1 queries showed forward hits enriched in the left arm and reverse-complement hits enriched in the right arm. Cross-arm sequence identity was broadly uniform (median 93.0%; median Jukes-Cantor distance = 0.0730), with nearly identical within-arm and cross-arm divergence distributions (median JC 0.0747 vs. 0.0735), providing no evidence for ongoing inter-arm gene conversion. Two block pairs showed substantially higher sequence identity than the main palindrome bulk (B03<-> B17, 98.6%; B06<-> B20, 97.1%), suggesting more recent localized duplication. Three-species synteny comparison identified a 7.1 Mb orientation-discordant interval, shared by mouse and rat relative to human, that encompasses SD12A1 and corresponds to a rearrangement-prone human orthologous region.</p><p><strong>Conclusions: </strong>These findings establish SD12A1 as a mouse-specific palindromic manifestation of an evolutionarily labile genomic interval and provide a reference-based framework for recognizing large autosomal palindromic SD architecture in mammalian genomes.</p>","PeriodicalId":9030,"journal":{"name":"BMC Genomics","volume":"27 1","pages":""},"PeriodicalIF":3.9,"publicationDate":"2026-07-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13688818/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149236915","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Integrative population genomics and tissue-specific expression profiling in cattle using whole-genome sequence resources.","authors":"Mohammed Saeed-Zidane, Iulia Blaj, Amira Yousif, Georg Thaller","doi":"10.1186/s12864-026-13218-4","DOIUrl":"10.1186/s12864-026-13218-4","url":null,"abstract":"<p><p>Large-scale genomic resources enable systematic investigation of genetic variation and tissue-specific expression of immune-related pathways in livestock species. Here, we integrated population-level whole-genome sequence data with tissue-specific transcriptional profiling to characterize genomic differentiation and expression patterns of NF-κB pathway-related genes in cattle. Using whole-genome sequence data from the 1000 Bull Genomes Project, we identified prioritized differentiated variants within several NF-κB pathway-related genes, including CD14, NF-κB1, IL-1β, and BAFFR. The mRNA expression levels of selected NF-κB signaling-related genes were quantified across four tissues with relevance to host defense and metabolism in bulls and cows of both breeds. Expression analyses revealed pronounced tissue-dependent regulation, with distinct organ-specific transcriptional patterns across the investigated genes. Intestinal tissues showed lower expression of several innate immune genes, whereas CD14 expression was more prominent in liver-associated comparisons, highlighting functional tissue specialization within the NF-κB signaling network. Breed- and sex-associated effects were gene-dependent rather than uniform across the pathway. Histological assessment of the spleen was performed to provide an anatomical context for observed transcriptional variations. A population-prioritized CD14 missense variant (Asn177Asp), identified from the 1000 Bull Genomes dataset, was contextualized using in silico structural annotation to provide structural context for coding variation. These findings provide descriptive evidence that population-level genomic differentiation in selected NF-κB pathway genes is accompanied by tissue-specific transcriptional differences in cattle.</p>","PeriodicalId":9030,"journal":{"name":"BMC Genomics","volume":"27 1","pages":""},"PeriodicalIF":3.9,"publicationDate":"2026-07-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13401305/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148590367","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"DNA methylation defines lineage identity yet converges among functionally related cells and is cell‑specifically remodeled during in vitro passaging.","authors":"Lanlan Sun, Dongmei Chen, Xin Zhang, Hailang Fan, Yaqian Zhao, Yingdi Huang, Yang Wu, Yanni Lou, Dake Zhang","doi":"10.1186/s12864-026-13224-6","DOIUrl":"10.1186/s12864-026-13224-6","url":null,"abstract":"<p><strong>Background: </strong>DNA methylation (DNAme) is used to define cellular identity, but its ability to distinguish closely related subpopulations and its stability during in vitro culture remain to be systematically evaluated.</p><p><strong>Results: </strong>Lineage-scale whole-genome bisulfite sequencing (WGBS) analysis of 40 cell types across 13 organs showed that DNAme separates major lineages but converges among functionally related populations. Endothelial cells, fibroblasts, and smooth muscle cells from different organs clustered primarily by lineage rather than tissue of origin, whereas epithelial cells retained stronger organ-associated methylation signatures. Lineage-restricted analysis of endothelial cells identified 700 tissue-specific differentially methylated regions (DMRs) across seven tissues that were masked in global comparisons, indicating residual tissue-of-origin information persists within highly similar lineages. In kidney cells, in vitro passaging induced systematic methylation remodeling without erasing lineage identity. Primary and subcultured glomerular endothelial and tubular epithelial cells formed separate subclusters within their lineage groups. Genome-wide analysis identified extensive differential methylation (~ 1.88 M differentially methylated CpG sites [DMCs] in glomerular endothelial; ~2.64 M in tubular epithelial), with promoters overrepresented among hypermethylated features. Shared hypermethylated promoters were enriched for core homeostatic functions; hypomethylated promoters were enriched for differentiation and tissue-specific pathways. Motif analysis showed enrichment of bZIP and GATA factor motifs in endothelial hypomethylated regions and ETS-family motifs in epithelial cells. Cross-cell-type array analysis confirmed reproducible methylation drift across fetal skin fibroblasts, adult vascular endothelial cells, and neonatal foreskin fibroblasts, with 7,434 shared hypomethylated DMCs and 1,347 shared hypermethylated DMCs. Drift onset was cell type-dependent: adult skin fibroblasts and vascular smooth muscle cells diverged by passage 3, whereas neonatal and fetal cells remained stable through passage 10. Shared hypomethylated DMCs were enriched in partially methylated domains (PMDs) and late-replicating domains (LRDs). Cross-platform comparison detected only 44 common hypomethylated DMCs and 24 common hypermethylated DMCs between array and WGBS data, with negligible regional overlap.</p><p><strong>Conclusions: </strong>DNAme robustly distinguishes major lineages but converges among closely related cell types, limiting fine subpopulation resolution. Lineage-restricted analysis recovers tissue-of-origin signals otherwise obscured, whereas in vitro passaging induces cell type-dependent remodeling and reproducible methylation drift that preserves lineage identity while attenuating tissue-specific programs. No universal passage threshold ensures phenotypic stability.</p>","PeriodicalId":9030,"journal":{"name":"BMC Genomics","volume":"27 1","pages":""},"PeriodicalIF":3.9,"publicationDate":"2026-07-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13685235/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149227103","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Interactions between kidney proteins and metabolism in Tibetan sheep at different altitudes.","authors":"Kejiang Liu, Jiang Zhang, Xueqin Yu, Pengfei Zhao","doi":"10.1186/s12864-026-13228-2","DOIUrl":"10.1186/s12864-026-13228-2","url":null,"abstract":"<p><p>When oxygen supply to the kidneys is restricted, it may trigger a series of acid-base imbalances and pathological states, subsequently inducing hypoxic adaptive mechanisms in the kidneys. However, the protein-metabolism interaction mechanisms underlying this process remain unclear. Therefore, this study conducted proteomic and metabolomic analyses on Tibetan sheep kidneys collected at three distinct altitudes (2500 m, 3500 m and 4500 m) to elucidate the protein-metabolism interaction mechanisms during hypoxic adaptation. The results demonstrate a synergistic relationship between proteins and metabolism, revealing dynamic expression patterns. Key DEPs (ALDH2, COX5A, GLUD1, NDUFS2, NDUFS8 and SLC17A1) and key DEMs (Wedelolactone, Cytidine, and 3-hydroxybenzoic acid) primarily participated in the antioxidant defense, anti-inflammatory, and energy supply functions of Tibetan sheep kidney hypoxia adaptation through signaling pathways such as the citrate cycle (TCA cycle) and oxidative phosphorylation. In summary, the adaptive mechanisms to renal hypoxia are not influenced by a single protein or metabolite, but rather are related to the interactions between proteins and metabolites. These interactions may collectively influence the adaptive mechanisms, ensuring that the kidneys maintain normal function in high-altitude environments.</p>","PeriodicalId":9030,"journal":{"name":"BMC Genomics","volume":"27 1","pages":""},"PeriodicalIF":3.9,"publicationDate":"2026-07-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13685116/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149227174","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Seasonal regulation of mRNAs and lncRNAs under cold exposure favours metabolic and immune-related pathways in Sahiwal Cattle.","authors":"Pradyut Das, Rani Alex, Varadanayakanahalli N Sahana, Pritam Pal, Kashif Dawood Khan, Amritanshu Upadhyay, Shambhavi Sharma, Gopal R Gowane, Goutam Mondal, Vikas Vohra","doi":"10.1186/s12864-026-13223-7","DOIUrl":"https://doi.org/10.1186/s12864-026-13223-7","url":null,"abstract":"<p><strong>Background: </strong>Seasonal fluctuations in environmental conditions impose physiological challenges on livestock, necessitating adaptive molecular responses. While heat-stress biology in cattle is well studied, responses to cold exposure remain largely unexplored in indigenous Indian breeds such as Sahiwal, one of the country's best milch cattle. To investigate seasonal adaptation mechanisms, we conducted transcriptome profiling of peripheral blood mononuclear cells (PBMCs) collected from Sahiwal cattle during winter and spring.</p><p><strong>Results: </strong>RNA-Seq analysis identified 7,659 differentially expressed mRNAs (DE-mRNAs; 1,332 upregulated and 6,327 downregulated) and 2,369 differentially expressed long non-coding RNAs (DE-lncRNAs; 106 upregulated and 2,263 downregulated), using thresholds of |log₂(fold change)| > 2 and false discovery rate < 0.05. Functional enrichment analysis of DE-mRNAs and cis-mRNAs of DE-lncRNAs using Gene Ontology, KEGG, Reactome pathways, and STRING networks showed that genes involved in cellular metabolism, immune responses, translation, and oxidative stress, such as MYC, CD4, PTPRC, MTSS1, CYBB, TLR10, THAP12, AIFM1, COX6A2, and SOD1, were upregulated during cold exposure. Conversely, genes associated with cell proliferation, response to external stimuli, tissue development, and angiogenesis, including IGF1, BDNF, SRC, YES1, CTTN, G0S2, CDK15, CCNA1, TLCD1, and BCL2L1, were downregulated.</p><p><strong>Conclusions: </strong>These patterns suggest an energy trade-off favouring thermoregulation and immune function over tissue development in winter. This study provides new molecular insights into seasonal physiological trade-offs in Sahiwal cattle, highlighting adaptive gene-expression changes that support cold resilience. The findings enhance our understanding of environmental adaptability in indigenous breeds and may aid in breeding climate-resilient livestock.</p>","PeriodicalId":9030,"journal":{"name":"BMC Genomics","volume":"27 1","pages":""},"PeriodicalIF":3.9,"publicationDate":"2026-07-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149227150","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"GenVarLoader: an accelerated dataloader for applying deep learning to personalized genomics.","authors":"David Laub, Aaron Ho, Loukik Raina, Jeff Jaureguy, Adam Klie, Rany M Salem, Graham McVicker, Hannah Carter","doi":"10.1186/s12864-026-13225-5","DOIUrl":"10.1186/s12864-026-13225-5","url":null,"abstract":"<p><strong>Background: </strong>Deep learning models can predict the functional impact of genetic variants. However, training and applying these sequence models to personalized genome sequences are severely constrained by the computational burden of storing and reading massive datasets.</p><p><strong>Results: </strong>We developed GenVarLoader, an accelerated dataloader that creates personalized genomic sequences and functional tracks on-the-fly to overcome these bottlenecks. GenVarLoader stores this personalized genomic data in formats specifically optimized for machine learning. Our tool reads variants up to 26,000 times faster than a pre-subset BCF, loads personalized data up to 130 times faster than tuned FASTA and pyBigWig pipelines, and reduces storage approximately 2,000-fold compared to existing alternatives.</p><p><strong>Conclusions: </strong>By eliminating severe data loading bottlenecks, GenVarLoader ensures efficient model training and inference. It provides a scalable solution for integrating biobank-scale personalized genomic data with deep learning applications.</p>","PeriodicalId":9030,"journal":{"name":"BMC Genomics","volume":"27 1","pages":""},"PeriodicalIF":3.9,"publicationDate":"2026-07-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13682130/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149223304","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Genome-wide identification and comparative analysis of endogenous retroviruses across five breed-representative pig assemblies.","authors":"Xiaotong Zhao, Yongping Hu, Zhidong Zhang, Xun Xiang, Hua Chang","doi":"10.1186/s12864-026-13211-x","DOIUrl":"10.1186/s12864-026-13211-x","url":null,"abstract":"&lt;p&gt;&lt;strong&gt;Background: &lt;/strong&gt;Porcine xenotransplantation offers a promising solution to human organ shortages, yet cross-species transmission of endogenous retroviruses (ERVs) remains a major safety barrier. Current risk assessments predominantly rely on a single reference genome, creating a distinct blind spot regarding ERV diversity and structural variation across diverse genetic backgrounds. We profiled ERV loci across five chromosome-level pig genome assemblies (Bama miniature pig, Duroc, Meishan, Ningxiang, and USMARC), characterizing structural completeness, phylogenetic relationships, relative insertion ages, and host-gene/QTL genomic context. ERVs were identified with the LTRharvest/LTRdigest pipeline, annotated by Pfam domain screening, and classified as complete, near-complete, or partial. Phylogenetic analysis used maximum-likelihood inference, insertion age was estimated from paired long terminal repeat (LTR) divergence, and ERV-proximal genes within 10 kb were analyzed for Gene Ontology (GO)/Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment and tissue expression.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Results: &lt;/strong&gt;Across the five chromosome-level assemblies, the paired-LTR-anchored catalog comprised 23,767 ERV candidates in the main autosome/chrX analysis set, including 75 structurally complete, 322 near-complete, and 23,370 partial loci. An additional 86 chrY records were detected in assemblies with chrY and were all classified as partial. Complete-ERV insertion-age profiles showed assembly-level variation, although Kruskal-Wallis tests did not detect a significant overall difference among assemblies. The five-assembly shared ERV-proximal host-gene set contained 278 genes at the 10 kb window; this set decreased to 229 genes at 5 kb and increased to 358 and 618 genes at 20 and 50 kb, respectively. Chromosome-stratified randomization showed that the shared 10 kb gene set overlapped pig QTL intervals more often than expected by chance (56.47% observed versus 43.23% random mean; empirical P &lt; 0.0001).&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Conclusions: &lt;/strong&gt;Across five breed-representative pig assemblies, we identified 23,767 paired-LTR ERV candidates, of which only 75 (0.32%) met the structural-completeness criteria, confirming that the detected porcine ERV catalog is dominated by degraded elements. Complete loci showed assembly-level variation in retroviral affinities and relative insertion-age profiles, while the five-assembly shared 10 kb ERV-proximal gene set showed non-random QTL colocalization (56.47% observed versus 43.23% random mean; empirical P &lt; 0.0001). Benchmarking against published porcine endogenous retrovirus (PERV) and retroviral insertion polymorphism (RIP) resources linked 9 complete loci to reconstructed Sus scrofa ERV6A/B full-length ERVs (SscERV6A/B Fl-ERVs) and left 66 complete loci outside that focused set. This catalog provides concrete ERV loci and host-gene contexts for donor-line evaluation and future population-scale, T2T-informed studi","PeriodicalId":9030,"journal":{"name":"BMC Genomics","volume":"27 1","pages":""},"PeriodicalIF":3.9,"publicationDate":"2026-07-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13679001/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149223317","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"First mitochondrial genomes in the Effusa clade (Gastropoda, Ampullariidae): unveiling new insights from Neotropical endemic Pomacea spp.","authors":"Alejandro Mendivil, Rina Ramírez, Jaime Morin, Jorge L Ramirez, Betty Shiga, Nilda Oliveros, Carlos Congrains, Michael D Martin","doi":"10.1186/s12864-026-13191-y","DOIUrl":"10.1186/s12864-026-13191-y","url":null,"abstract":"<p><strong>Background: </strong>Pomacea is the most diverse genus of the freshwater snail family Ampullariidae, which have been proposed as model organisms in biogeography, speciation and invasive biology. While substantial effort has been put into the mitochondrial genomes of Pomacea clades associated with invasive species (P. canaliculata and P. bridgesii groups), comparatively little is known about the mitogenomic organization and structure within the Effusa clade.</p><p><strong>Results: </strong>This study provides the first complete and annotated mitochondrial genomes of the Effusa clade, based on Pomacea nobilis (Reeve, 1856) and a putative undescribed Pomacea lineage (hereafter Pomacea sp.). The mitochondrial genome of P. nobilis (16,279 - 16,471 bp) and Pomacea sp. (16,355 bp) are among the largest recorded for Pomacea and contain the typical 37 animal genes. A comparative analysis with other Ampullariidae revealed that the Effusa clade differs from both P. canaliculata and P. bridgesii clades mainly by the higher proportion of intergenic regions, the size of many genes (ATP8, ATP6, NAD6, COIII and NAD2), the largest ribosomal RNAs among Pomacea and a putative large repetitive unit in the control region.</p><p><strong>Conclusion: </strong>These findings highlight the distinctiveness of Effusa clade mitogenomes in comparison with the P. canaliculata and P. bridgesii groups. Further mitogenomic studies on underexplored Pomacea lineages will be important to broaden our understanding of mitochondrial genome diversity among apple snails.</p>","PeriodicalId":9030,"journal":{"name":"BMC Genomics","volume":"27 1","pages":""},"PeriodicalIF":3.9,"publicationDate":"2026-07-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13681941/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149223257","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Draft hybrid genome assembly of the photosynthetic sacoglossan Elysia crispata provides insights into sacoglossan biology.","authors":"Padmanabhan Mahadevan, Michael L Middlebrooks, Nicholas E Curtis","doi":"10.1186/s12864-026-13157-0","DOIUrl":"10.1186/s12864-026-13157-0","url":null,"abstract":"<p><p>We present here a draft genome of the photosynthetic sacoglossan sea slug, Elysia crispata (ecotype crispata) generated with a reference guided approach using Illumina short reads and long PacBio HiFi reads. The genome size is approximately 671 Mb with an N50 of 460 kb. BUSCO genome completeness was 98.2%, comprising 97.5% single-copy complete and 0.7% duplicated complete metazoan core genes, with an additional 0.7% partial core genes. This genome provided interesting observations such as the possible remnants of an endogenous retrovirus and the presence of polyketide synthesis proteins thought to protect the slug from photosynthetic damage, based on homology with characterized sacoglossan PKS proteins. Few genomes of photosynthetic sacoglossans are available and this draft genome of the E. crispata sea slug will serve as a valuable dataset to study sacoglossan biology, genetics, evolution, and photosymbiosis. In addition, our assembly strategy may serve as a guide to other researchers who also do not have Hi-C data to assemble their genomes, though this approach relies on the availability of a closely related reference genome for scaffolding.</p>","PeriodicalId":9030,"journal":{"name":"BMC Genomics","volume":"27 1","pages":""},"PeriodicalIF":3.9,"publicationDate":"2026-07-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13673630/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149206554","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Systematic evaluation of single-cell foundation model interpretability: attention-derived edge scores add no incremental value over gene-level features for perturbation-target prediction.","authors":"Ihor Kendiukhov","doi":"10.1186/s12864-026-12965-8","DOIUrl":"10.1186/s12864-026-12965-8","url":null,"abstract":"&lt;p&gt;&lt;strong&gt;Background: &lt;/strong&gt;Single-cell foundation models such as scGPT and Geneformer are increasingly used for gene regulatory network (GRN) inference, with attention-derived edge scores routinely interpreted as regulatory proxies. Prior benchmarks have evaluated curated-reference recovery but have not systematically tested whether attention adds information beyond expression statistics for predicting the outcomes of genetic perturbations, nor whether attention-identified \"regulatory\" components are causally required for such predictions. This gap matters because the NLP interpretability literature has established that attention weights do not reliably indicate feature importance, and biological foundation models are being deployed without analogous scrutiny.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Methods: &lt;/strong&gt;We present an evaluation framework comprising thirty-seven analyses and 153 statistical tests under Benjamini-Hochberg FDR correction, spanning two architectures (scGPT, Geneformer V2-316M), four cell types (K562, RPE1, primary T cells, iPSC neurons), and two perturbation modalities (CRISPRi, CRISPRa). The framework separates two objectives: (A) mechanistic interpretability / GRN recovery against curated references, and (B) perturbation-target prediction, i.e. classifying which genes show differential expression after a CRISPR perturbation. Five test families-trivial-baseline comparison, conditional incremental-value testing, residualisation and propensity matching, causal ablation with intervention-fidelity diagnostics, and cross-context replication-address Objective B, supplemented by a synthetic positive control establishing pipeline sensitivity.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Results: &lt;/strong&gt;Attention patterns encode layer-specific biological structure-protein-protein interactions in early layers, transcriptional regulation in late layers-and Cell-State Stratified Interpretability (CSSI) exploits this structure to improve curated GRN recovery up to [Formula: see text] on the Objective A task. On Objective B, however, attention-derived edge scores add no incremental value beyond trivial gene-level features (variance, mean expression, dropout rate): gene-level baselines outperform both attention and correlation edges (AUROC 0.81-0.88 versus 0.70), augmenting gene-level predictors with pairwise edges produces ΔAUROC of [Formula: see text] to [Formula: see text] across 559,720 perturbation-gene observations, and causal ablation of TRRUST-ranked attention heads produces no degradation across three independent intervention channels. The attention-correlation relationship is context-dependent (equal in K562 CRISPRi, worse in CRISPRa, better in RPE1), but gene-level dominance on Objective B is universal across both cell types where adequate power is available.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Conclusions: &lt;/strong&gt;Attention patterns in single-cell foundation models encode biologically structured information, including layer-specific regulatory signals recoverable via CSSI, but provide no uni","PeriodicalId":9030,"journal":{"name":"BMC Genomics","volume":" ","pages":""},"PeriodicalIF":3.9,"publicationDate":"2026-07-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13390336/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148547753","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}