High-quality mouse reference genomes reveal the structural complexity of the murine protein-coding landscape

Initial short-read assemblies of mouse strains are incomplete in many complex and repeat-rich regions.

“Of note, these initial assemblies, produced using short-read sequencing, are incomplete in many complex and repeat-rich regions.”

• cites [18] Lilue, 2018, Sixteen diverse laboratory mouse reference genomes define strain-specific haplotypes and novel functional loci, Nat. Genet. sourcePartly supported

  Online Methods (De novo assembly; page unavailable) says the initial contigs were assembled from paired-end reads. Results (Sequence assemblies and genome annotation; page unavailable) reports that long-repeat sequence lengths were consistently shorter than in the reference, supporting incomplete representation of repeats. The copy does not establish that many complex regions in these initial strain assemblies were incomplete. (LLM, conf=0.82)

  Experimental — accuracy not yet measured

The cumulative size of non-reference regions per chromosome varies between 1.26 Kbp per 1 Mbp in chromosome 16 of C57BL/6NJ and 175.82 Kbp per 1 Mbp of chromosome X in SPRET/EiJ.

“Figure 1C shows the cumulative size of these non-reference regions per chromosome, which varies between 1.26 Kbp per 1 Mbp in chromosome 16 of C57BL/6NJ and 175.82 Kbp per 1 Mbp of chromosome X in SPRET/EiJ.20”

• cites [20] Mahler, 2008, Sequence divergence of Mus spretus and Mus musculus across a skin cancer susceptibility locus, BMC Genom. sourceCould not verify

  The Results describe sequencing at the Skts5 locus on chromosome 12, and the Discussion says the study was limited to that locus. Searches for the stated values and for non-reference regions on chromosomes 16 and X found no corresponding measurements in our copy, so it cannot verify the chromosome-wide range. (LLM, conf=0.77)

  Experimental — accuracy not yet measured

Evolutionary gain and loss of Gbp family members have shaped the diverse immune responses in different mouse strains.

“Evolutionary gain and loss of Gbp family members have shaped the diverse immune responses in different mouse strains.22,23”

• cites [22] Côrte-Real, 2022, Evolution of Guanylate Binding Protein (GBP) Genes in Muroid Rodents (Muridae and Cricetidae) Reveals an Outstanding Pattern of Gain and Loss, Front. Immunol. sourcePartly supported

  The Discussion (page not provided) reports gains and losses of Gbp genes across muroid species and suggests that this diversity is consistent with their role in host defense. The Introduction describes immune functions of mouse GBPs. The paper does not establish that evolutionary differences among mouse strains caused their differing immune responses; its evolutionary comparison is primarily across species. (LLM, conf=0.86)

  Experimental — accuracy not yet measured
• cites [23] Clough, 2019, C57BL/6 and 129 inbred mouse strains differ in Gbp2 and Gbp2b expression in response to inflammatory stimuli in vivo, Wellcome Open Res. sourcePartly supported

  The Results identify a Gbp2b/Gbp2 gene duplication in 129/Sv mice, and the Discussion reports strain differences in Gbp2b and Gbp2 expression after immune stimulation (no page numbers provided). These findings support a connection between gene-copy variation and inflammatory gene expression, but the paper does not establish evolutionary gain and loss of Gbp family members as the cause of diverse immune responses across mouse strains. (LLM, conf=0.84)

  Experimental — accuracy not yet measured

In 2019, the authors determined the most genetically diverse regions of the mouse genome, where genetic variation can exceed interspecies variation.

“In 2019, we determined the most genetically diverse regions of the mouse genome, where genetic variation can exceed interspecies variation.18,19”

• cites [18] Lilue, 2018, Sixteen diverse laboratory mouse reference genomes define strain-specific haplotypes and novel functional loci, Nat. Genet. sourcePartly supported

  The Results section, “Regions of the mouse genome with extreme allelic variation” (page unavailable), identifies the most hSNP-dense regions, supporting the claim that the authors determined highly diverse regions. The Introduction says that differences between mouse strains at some loci are comparable to those between humans and chimpanzees, not that they can exceed interspecies variation. The cited paper is dated 2018, not 2019. (LLM, conf=0.87)

  Experimental — accuracy not yet measured
• cites [19] Lilue, 2019, Mouse protein coding diversity: What’s left to discover?, PLoS Genet. sourcePartly supported

  The Conclusions identify strain-specific diversity regions (SSDRs) as unusually variable, with differences between mouse strains often exceeding mouse–rat diversity. But the Limitations of a single mouse reference genome section credits the identification of 2,567 SSDRs to a 2018 study; this 2019 paper reviews that finding rather than presenting it as a new determination. (LLM, conf=0.85)

  Experimental — accuracy not yet measured

Using long-read assemblies as reference genomes improves alignment metrics compared to GRCm39 or short-read assemblies.

“We found improvement across several alignment metrics (Figure 1B; Table S3) when using long-read assemblies compared to GRCm39 or the short-read sequencing assemblies.18”

• cites [18] Lilue, 2018, Sixteen diverse laboratory mouse reference genomes define strain-specific haplotypes and novel functional loci, Nat. Genet. sourceSupported

In 2002, the mouse reference genome was based on a single inbred mouse strain, C57BL/6J (BL6).

“In 2002, the mouse reference genome was based on a single inbred mouse strain, namely, C57BL/6J (BL6).3”

• cites [3] Waterston, 2002, Initial sequencing and comparative analysis of the mouse genome, Nature sourceSupported

Until now, most gene expression studies of different mouse strains have used the BL6 reference genome despite millions of known genetic variations between strains.

“Until now, most gene expression studies of different mouse strains have used the BL6 reference genome despite millions of known genetic variations between strains.14,15”

• cites [14] Doran, 2016, Deep genome sequencing and variation analysis of 13 inbred mouse strains defines candidate phenotypic alleles, private variation and homozygous truncating mutations, Genome Biol. sourcePartly supported

  The Background identifies C57BL/6J as the basis of the mouse reference genome, and the Results report millions of variants across other strains. The Methods describe aligning this paper’s DNA-sequencing reads to that reference, but no passage establishes that most gene-expression studies of different strains used it (searched for gene-expression studies and RNA-seq mapping). (LLM, conf=0.86)

  Experimental — accuracy not yet measured
• cites [15] Keane, 2011, Mouse genomic variation and its effect on phenotypes and gene regulation, Nature sourcePartly supported

  The Abstract identifies C57BL/6J as the mouse reference genome, and Data generation and variant discovery (page not provided) reports millions of variants across strains. Neither passage establishes that most gene-expression studies of different strains used that reference genome; the paper describes its own analysis, not the prevalence of that practice. (LLM, conf=0.82)

  Experimental — accuracy not yet measured

The de novo assembly was validated with plasmid and fosmid results.

“To confirm the accuracy of our assembly, we validated it with plasmid and fosmid results.33,34,35,36,37,38”

• cites [33] Weiss, 1984, Organization and evolution of the class I gene family in the major histocompatibility complex of the C57BL/10 mouse, Nature sourceNot checked — no full text of the source
• cites [34] Watts, 1989, Organization and structure of the Qa genes of the major histocompatibility complex of the C3H mouse: implications for Qa function and class I evolution, EMBO J. sourceCould not verify

  The Materials and methods and DNA sequencing sections describe cloning, restriction mapping, and sequencing C3H mouse Qa genes. A search for de novo assembly validation and plasmid/fosmid results found no passage addressing the claimed validation; the paper describes cosmid clones, not fosmid results. Page numbers were not available in our copy. (LLM, conf=0.76)

  Experimental — accuracy not yet measured
• cites [35] Rubocki, 1990, Molecular evidence that the H-2D and H-2L genes arose by duplication. Differences between the evolution of the class I genes in mice and humans, J. Exp. Med. sourceCould not verify

  The DNA Sequencing methods describe subcloning a gene fragment into a plasmid vector, and the mapping methods describe cosmid clones (pp. 2045–2046). Neither passage reports fosmid results or validation of a de novo assembly; searches for those terms found no supporting passage in our copy. (LLM, conf=0.76)

  Experimental — accuracy not yet measured
• cites [36] Brown, 1988, Extension of the H-2 TLb molecular map. Isolation and characterization of T13, T14, and T15 from the C57BL/6 mouse, Immunogenetics sourceNot checked — no full text of the source
• cites [37] Fisher, 1985, Structure of a gene encoding a murine thymus leukemia antigen, and organization of Tla genes in the BALB/c mouse, J. Exp. Med. sourceCould not verify

  The “New Linkage of Tla Clusters” passage (pp. 539–540) describes checking cosmid-based gene-cluster maps against Southern blots, not validating a de novo assembly with plasmid and fosmid results. A search of our copy for fosmids and assembly validation found no passage addressing that claim. (LLM, conf=0.76)

  Experimental — accuracy not yet measured
• cites [38] Litaker, 1996, A physical map of the Q region of B10.P. Mamm, Genome sourceNot checked — no full text of the source

High sequence diversity within CDS and a high dN/dS value indicate directional selection (positive selection) for H2-K and classical class II MHC molecules.

“However, the high sequence diversity within protein-coding sequence (CDS) and a high dN/dS value indicate directional selection (positive selection), as previously reported.31,32”

• cites [31] Radwan, 2020, Advances in the Evolutionary Understanding of MHC Polymorphism, Trends Genet. sourcePartly supported

  The Abstract (page unavailable) describes extreme polymorphism in classical MHC genes and says a high relative nonsynonymous substitution rate in the peptide-binding domain is evidence of adaptive evolution. The Reconciling Landmark Features of MHC section (page unavailable) likewise links excess nonsynonymous variation to positive selection. Our copy does not establish a high dN/dS value or that this finding applies specifically to H2-K and classical class II molecules. (LLM, conf=0.84)

  Experimental — accuracy not yet measured
• cites [32] Bill, 1989, Positive selection of CD4+ T cells mediated by MHC class II-bearing stromal cell in the thymic cortex, Nature sourceNot checked — no full text of the source

The killer cell lectin-like receptor locus (Klra; Ly49) in mice is another important gene family with significant numbers of non-reference genes, such as Ly49L (Klra12) in BALB or Ly49T (Klra20) in 129-related strains.

“The killer cell lectin-like receptor locus (Klra; also called the Ly49 locus) in mice is another important gene family with significant numbers of non-reference genes, such as Ly49L (Klra12) in BALB or Ly49T (Klra20) in 129-related strains.49”

• cites [49] Rahim, 2014, Ly49 receptors: innate and adaptive immune paradigms, Front. Immunol. sourcePartly supported

  The Abstract identifies the mouse Ly49 gene family as Klra. Table 1 lists Ly49L in BALB and Ly49T in 129 but not in B6, and the Genetic diversity section describes substantial differences in gene content among strains. Our copy does not give the Klra12 or Klra20 aliases or quantify how many genes are non-reference. (LLM, conf=0.86)

  Experimental — accuracy not yet measured

Different copy numbers of repeats between C57BL/6NJ and GRCm39 in Ubc, Ahnak2, Flg, Gm5154, and B230307C23Rik were confirmed as GRCm39 assembly mistakes by cross-checking with BL6 T2T assemblies.

“we identified different copy numbers of repeats between C57BL/6NJ and GRCm39 in the sequence of Ubc, Ahnak2, Flg, Gm5154, and B230307C23Rik and then confirmed them as GRCm39 assembly mistakes by cross-checking them with the BL6 T2T assemblies.29,30”

• cites [29] Francis, 2024, The structural diversity of telomeres and centromeres across mouse subspecies revealed by complete assemblies, Genetics sourcePartly supported

  The Chromosome Structure and annotation section reports gene copy-number differences between GRCm39 and the T2T C57BL/6J assembly. It does not identify repeat-copy differences in Ubc, Ahnak2, Flg, Gm5154, or B230307C23Rik, compare C57BL/6NJ, or confirm mistakes at those loci in GRCm39. (LLM, conf=0.82)

  Experimental — accuracy not yet measured
• cites [30] Liu, 2024, The complete telomere-to-telomere sequence of a mouse genome, Science sourceNot checked — no full text of the source

The 6th exon of the mouse gene Ifi207 encodes 11-25 copies of an exact 42 bp repeat, which have accumulated extremely high diversity (15% at the nucleotide level and 29% at the amino acid level) within a few million years.

“Recently, we reported that the 6th exon of the mouse gene Ifi207 encodes 11–25 copies of an exact 42 bp repeat, which have accumulated extremely high diversity (15% at the nucleotide level and 29% at the amino acid level) within a few million years.55”

• cites [55] Moran, 2024, IFI207, a young and fast-evolving protein, controls retroviral replication via the STING pathway, mBio sourcePartly supported

  The Results section (“Ifi207 is characterized by two highly polymorphic repeat regions”; page unavailable) reports 11–25 repeat units of precisely 42 base pairs each, with average nucleotide and amino-acid diversity of 15% and 29%, respectively. It dates their emergence to approximately 6.5 million years ago. Our copy says the repeats are encoded in a single large exon, but does not identify it as the sixth exon. (LLM, conf=0.86)

  Experimental — accuracy not yet measured

Compared to previous de novo mouse strain genomes and variation catalogs, these long-read genome assemblies show significant improvement in the most complex regions, such as MHC, defensin, TCR, and Ly49 loci.

“Compared to previous de novo mouse strain genomes and variation catalogs,14,15,18 these long-read genome assemblies show significant improvement in the most complex regions, such as MHC, defensin, T cell receptor (TCR), and Ly49 loci.”

• cites [14] Doran, 2016, Deep genome sequencing and variation analysis of 13 inbred mouse strains defines candidate phenotypic alleles, private variation and homozygous truncating mutations, Genome Biol. sourceSupported
• cites [15] Keane, 2011, Mouse genomic variation and its effect on phenotypes and gene regulation, Nature sourceSupported
• cites [18] Lilue, 2018, Sixteen diverse laboratory mouse reference genomes define strain-specific haplotypes and novel functional loci, Nat. Genet. sourceSupported

Many of the newly discovered VNTR genes play important roles in human disease.

“Many of the newly discovered VNTR genes play important roles in human disease, e.g., STAT2 in antiviral immunity,64 HRNR in atopic dermatitis skin lesions65 and hepatocellular carcinoma,66 SBSN in tumor progression,67 KMT2D in Kabuki syndrome,68 FLG2 in peeling skin syndrome,69 KRTAPs in ectodermal dysplasia,70 MAGEB4 in azoospermia,71 and RAD18 in Alzheimer’s disease.72”

• cites [64] Bucciol, 2024, Spotlight: “Human STAT2 deficiency: a severe defect of antiviral immunity.”, Genes Immun. sourceNot checked — no full text of the source
• cites [65] Makino, 2024, Expression of hornerin in skin lesions of atopic dermatitis and skin diseases, Clin. Exp. Dermatol. sourcePartly supported

  The Results report HRNR expression in chronic atopic dermatitis lesions but not in the three acute lesions studied. The Introduction says HRNR’s function remains unknown, so the paper supports an association with atopic dermatitis lesions, not an established important role in the disease. (LLM, conf=0.83)

  Experimental — accuracy not yet measured
• cites [66] Fu, 2018, Hornerin promotes tumor progression and is associated with poor prognosis in hepatocellular carcinoma, BMC Cancer sourceSupported

The majority of the Apol gene family on chr15 is significantly upregulated in mouse lungs following Mtb infection.

“Previous transcriptomic studies of infected inbred BL6 and C3HeB/FeJ mice62 and outbred DO mice63 reveal that the majority of the Apol gene family on chr15 is significantly upregulated in mouse lungs following Mtb infection.”

• cites [62] Moreira-Teixeira, 2020, Mouse transcriptome reveals potential signatures of protection and pathogenesis in human tuberculosis, Nat. Immunol. sourceCould not verify

  The Results passages report changes in gene-expression modules in infected mouse lungs, but do not identify Apol genes or chromosome 15. Searches of our copy for “Apol” and “Apol gene family chr15” found no passage establishing that a majority of those genes were significantly upregulated. (LLM, conf=0.70)

  Experimental — accuracy not yet measured
• cites [63] Ahmed, 2020, Immune correlates of tuberculosis disease and risk translate across species, Sci. Transl. Med. sourceCould not verify

  The Results sections report gene-expression changes in Mtb-infected DO mice, but do not identify upregulation of the mouse Apol gene family or place it on chromosome 15. Searches for “Apol,” “chromosome 15,” and the claimed expression pattern found only a statement that human APOL1 lacks a mouse ortholog; that does not resolve the claim about other mouse Apol genes. No page number is available in our copy. (LLM, conf=0.72)

  Experimental — accuracy not yet measured

Among wild-derived strains, the fraction of reads where pairs align to different chromosomes varies between 0.58% and 3.00% using long-read assemblies, compared to 1.24%-4.67% using GRCm39 and 7.93%-9.88% using short-read assemblies.

“For instance, among the wild-derived strains, the fraction of reads where pairs align to different chromosomes varies between 0.58% and 3.00% using the long-read assemblies compared to 1.24%–4.67% and 7.93%–9.88% while using GRCm39 or the short-read assemblies, respectively.”

Among classical mouse strains, the non-reference sequence per chromosome spans from 12.8 to 58.57 Kbp per Mbp, with a median of 30.3 Kbp per Mbp.

“Among classical mouse strains, the non-reference sequence per chromosome spans from 12.8 to 58.57 Kbp per Mbp, with a median of 30.3 Kbp per Mbp.”

The scaffold N50s of several new assemblies are comparable to GRCm39 (11-116 Mbp vs. 100.9 Mbp).

“however, the scaffold N50s are comparable to those of GRCm39 for several of the genomes (11–116 Mbp vs. 100.9 Mbp for GRCm39).”

Strain-specific annotations improve RNA-seq read mapping by 5.1% in PWK/PhJ compared to GRCm39.

“Our strain-specific annotations enhance RNA sequencing (RNA-seq) analyses, as demonstrated in PWK/PhJ, where we observed a 5.1% improvement in read mapping and expression-level differences in 2.1% of coding genes compared to using GRCm39.”

A region on chromosome 5 (105.08-105.64 Mbp in GRCm39) encodes guanylate-binding proteins (GBPs).

“Among the loci of highest divergence in the pangenome (Figure 1A) is a region on chromosome 5 (105.08–105.64 Mbp in GRCm39) that encodes guanylate-binding proteins (GBPs), a group of large GTPases that play essential roles in immune responses, cell signaling, and host defense against pathogens.”

The de novo assemblies have total lengths for chromosomes 1-19 and X (excluding Ns) between 2.46 and 2.68 Gbp, compared to 2.56 Gbp in GRCm39.

“yielding assemblies with a total length for chromosomes 1–19 and X (excluding unknown bases, e.g., Ns) between 2.46 and 2.68 Gbp, compared to 2.56 Gbp in GRCm39.”

The total number of protein-coding genes ranges from 21,592 to 23,187 across the strains.

“The total number of protein-coding genes ranges from 21,592 to 23,187 across the strains.”

Assembly QV scores range between Q40 and Q52.18 (99.99%-99.9993% accuracy).

“with assembly QV scores ranging between Q40 and Q52.18 (99.99%–99.9993% accuracy).”

Strain-specific annotations reveal expression-level differences in 2.1% of coding genes in PWK/PhJ compared to GRCm39.

“Our strain-specific annotations enhance RNA sequencing (RNA-seq) analyses, as demonstrated in PWK/PhJ, where we observed a 5.1% improvement in read mapping and expression-level differences in 2.1% of coding genes compared to using GRCm39.”

In the current assemblies, these highly diverse regions are highly contiguous (96%-100% are spanned by less than three contigs).

“In our current assemblies, these regions are highly contiguous (96%–100% are spanned by less than three contigs; Table S2).”

The size distribution of non-reference regions shows an enrichment at 6-7 Kbp, reflecting the size of full-length LINEs.

“Notably, the size distribution shows an enrichment at 6–7 Kbp, reflecting the size of full-length LINEs (Figure 1D).”

The genome assemblies achieve comparable sequence and gene annotation quality to the GRCm39 reference genome.

“These genome assemblies achieve comparable sequence and gene annotation quality to the GRm39 reference genome.”

The non-reference regions in the pangenome vary in size between 50 bp and 207.2 Kbp.

“The non-reference regions in the pangenome vary in size between 50 bp (default minimum threshold of minigraph) and 207.2 Kbp.”

Hundreds of non-reference genes from previous publications not found in GRCm39, such as Defa1, Raet1a, and Klra20 (Ly49T), were localized in the new reference genomes.

“Hundreds of non-reference genes from previous publications not found in GRCm39, such as Defa1, Raet1a, and Klra20 (Ly49T), were localized in the new reference genomes.”

The assembly base error rate ranges from 0.38 to 2.45 SNPs per 1 Mbp, compared to 2.07 for GRCm39.

“The assembly base error rate, estimated by aligning short reads from each strain to its respective assembly and calling SNPs, ranged from 0.38 to 2.45 SNPs per 1 Mbp, compared to 2.07 for GRCm39.”

The study resolved previously incomplete genomic regions, including the MHC, defensin cluster, T cell receptor, and Ly49 complexes.

“We have resolved previously incomplete genomic regions, including the major histocompatibility complex (MHC), defensin cluster, T cell receptor, and Ly49 complexes.”

The contig N50s of the new assemblies are lower than GRCm39 (0.82-33.9 Mbp vs. 57.4 Mbp).

“The contig N50s are lower than those of GRCm39 (0.82–33.9 Mbp vs. 57.4 Mbp for GRCm39);”

Genes in the loci of highest divergence are enriched for defense and immunity protein classes.

“We examined the protein classes of these genes using PANTHER, and the results show an enrichment for defense and immunity protein classes, highlighting the biomedical importance of such loci (Figure 1E).”

The genome-wide scan identified over 400 genes with VNTR polymorphisms in coding regions.

“We conducted a genome-wide scan of variable number tandem repeats (VNTRs) within the coding regions, identifying over 400 genes with VNTR polymorphisms with up to 600 repeat copies and repeat units reaching 990 nucleotides.”

The fraction of RNA-seq reads that align to CDS is 2%-10% higher across different tissues when aligned to the PWK/PhJ reference compared to GRCm39.

“The fraction of RNA-seq reads that align to CDS is 2%–10% higher across different tissues when aligned to the PWK/PhJ reference, compared to GRCm39 (Figure 4A).”

In haplotypes q and z, four novel H2-D homologs (D5, D6, D7, and D8) and three H2-Q (Q16, Q17, and Q18) were identified.

“In haplotypes q and z, we identified four novel H2-D homologs (D5, D6, D7, and D8) and three H2-Q (Q16, Q17, and Q18).”

VNTR polymorphisms were detected in 428 genes, around 3% of the total protein-coding genes in mice.

“VNTR polymorphisms were detected in 428 genes, around 3% of the total protein-coding genes in mice.”

Wild-derived strains show extensive diversity at the Gbp locus, with locus size varying greatly (0.39-0.48 Mbp) due to structural variants.

“Wild-derived strains show extensive diversity, with locus size varying greatly (0.39–0.48 Mbp) due to structural variants (including inversions, translocations, and duplications).”

The α-defensin locus in DBA/2J and C3H/HeJ share the same haplotype.

“We have analyzed the α-defensin locus within the de novo assemblies of DBA/2J and C3H/HeJ (Figure 2C), which share the same haplotype of α-defensin.”

Prior to the mouse reference genome, BL6 was not the most popular mouse strain for biomedical research.

“Prior to the mouse reference genome, BL6 was not the most popular mouse strain for biomedical research (Figure S1).”

In the genome-wide scan, variable VNTR examples with different evolutionary profiles were found.

“In our genome-wide scan, we found variable VNTR examples with different evolutionary profiles.”

Defa2 in the reference genome is not the ortholog of Defa2 cloned from C3H in previous research.

“Notably, Defa2 in the reference genome is not the ortholog of Defa2 cloned from C3H in previous research.42”

• cites [42] Huttner, 1994, A Family of Defensin-like Genes Codes for Diverse Cysteine-Rich Peptides in Mouse Paneth Cells, Genomics sourceNot checked — no full text of the source

None of the wild-derived strains share the same H2 haplotypes with laboratory strains or with each other.

“Notably, none of these wild-derived strains share the same H2 haplotypes with laboratory strains or with each other (Figure S4).”

The 17 genome assemblies have resolved H2 haplotypes a, bc, d, k, q, g7, and z in classical laboratory mouse strains, as well as H2(b) in C57BL/6NJ.

“Our 17 genome assemblies have resolved H2 haplotypes a, bc, d, k, q, g7, and z in the classical laboratory mouse strains, as well as the H2(b) in C57BL/6NJ.”

• AssumptionRests on an assumption: The mouse MHC H2 locus on chr17 is one of the most polymorphic regions in the mouse genome.

Highly conserved repeating sequences were found (e.g., Ubc, 8-23 copies of 228 bp repeats, Tajima's π = 8.95 × 10−3).

“Also, we found highly conserved repeating sequences (e.g., Ubc, 8–23 copies of 228 bp repeats, Tajima’s π = 8.95 × 10−3).”

The longest repeat unit is found in Eppk1 (984-990 bp).

“The longest repeat unit is found in Eppk1 (984–990 bp).”

For H2-K and classical class II MHC molecules, gene order and structure are highly conserved.

“For H2-K and classical class II MHC molecules, gene order and structure are highly conserved.”

Homologs of many published cryptdin-related sequence (CRS) peptides are absent in the GRCm39 reference genome.

“There are homologs of many published cryptdin-related sequence (CRS) peptides42 that are absent in the GRCm39 reference genome.”

• cites [42] Huttner, 1994, A Family of Defensin-like Genes Codes for Diverse Cysteine-Rich Peptides in Mouse Paneth Cells, Genomics sourceNot checked — no full text of the source

Gaps in H2-Q and H2-T loci of the GRCm39 reference genome have been filled.

“Gaps in H2-Q and H2-T loci of the GRCm39 reference genome have been filled.”

In GRCm39, 39 Defa members (including 9 pseudogenes) are encoded contiguously in a ∼800 Kbp locus at chr8:21.4-22.2 Mbp.

“In the GRCm39 reference, 39 Defa members (including 9 pseudogenes) are encoded contiguously in a ∼800 Kbp locus at chr8:21.4–22.2 Mbp on GRCm39.”

13 genes previously sequenced and published are absent in GRCm39, including Defa1, Defa4, Defa6, Defa7, and Defa14.

“We identified 13 genes previously sequenced and published that are absent in GRCm39 (shown in green), including Defa1, Defa4, Defa6, Defa7, and Defa14.”

Five additional H2 haplotypes distinct from any known laboratory mouse H2 haplotype were identified in five wild-derived strains.

“We performed the same analysis on five wild-derived strains (WSB/EiJ, PWK/PhJ, JF1/MsJ, CAST/EiJ, and SPRET/EiJ), identifying five additional haplotypes that are distinct from any known laboratory mouse H2 haplotype (see STAR Methods).”

359 genes (2.1% of 16,222 genes with clear 1-to-1 paralogs) are either differentially expressed in the opposite direction or change to a differential expression state (or vice versa) when comparing PWK/PhJ reference to GRCm39.

“We identified 359 genes (2.1% of 16,222 genes with clear 1-to-1 paralogs) that are either differentially expressed in the opposite direction or change to a differential expression state (or vice versa) (Figures 4B and S8).”

Several gene families are enriched with VNTRs, including Krab-Zfp, keratins, keratin-related proteins, and mucins.

“Several gene families are enriched with VNTRs, for example, Krab-Zfp members, keratins (Krt members), keratin-related proteins (Krtap and Tspear family), and mucins (Muc family).”

Only 19 Defa genes have direct orthologs in the reference.

“However, only 19 genes have direct orthologs in the reference (shown in blue).”

Five different haplotypes of the mouse Ly49 (Klra) locus from 12 laboratory mouse strains were fully assembled.

“We have fully assembled five different haplotypes of the mouse Ly49 (Klra) locus from 12 laboratory mouse strains (Figure S7).”

Other previously cloned genes (e.g., Defa8, Defa9, Crs4c1b) cannot be found in C3H/HeJ or DBA/2J.

“Other previously cloned genes (e.g., Defa8, Defa9, Crs4c1b, etc.) cannot be found in our sequenced strains C3H/HeJ or DBA/2J.”

278 gene markers absent from GRCm39 were identified from MGI Entrez gene markers.

“We downloaded the Entrez gene markers from the Mouse Genome Informatics (MGI) database (https://www.informatics.jax.org/downloads/reports/MGI_EntrezGene.rpt) and identified 278 gene markers absent from GRCm39 (“non-withdrawn” and without “genome coordinate”).”

Apol9a and Apol9b reads are misaligned due to high-density SNPs in the coding sequence.

“Apol9a and Apol9b (red), where there is a misalignment of reads caused by high-density SNPs in the coding sequence”

Across the CC/DO founders, a non-syntenic region of approximately 0.4 Mbp in length was identified on chr15, containing an inversion present in BL6 and 129S1/SvImJ and not found in the remaining seven founder lines.

“Across the CC/DO founders, a non-syntenic region of approximately 0.4 Mbp in length was identified on chr15, containing an inversion present in BL6 and 129S1/SvImJ and not found in the remaining seven founder lines (Figure 5E).”

Each of the non-reference founders harbors 6-8 de novo protein-coding genes within the non-syntenic region on chr15.

“Leveraging novel RNA-seq-based gene annotations, each of the non-reference founders was found to harbor 6–8 de novo protein-coding genes within this non-syntenic region (Figure 5E).”

Our data indicate that VNTRs in protein-coding genes are not caused by a single mechanism.

“However, our data indicate that VNTRs in protein-coding genes are not caused by a single mechanism.”

In a BMDM infection model, CAST/EiJ macrophages exhibited earlier cell death than BL6 macrophages following infection.

“In a bone-marrow-derived macrophage (BMDM) infection model, CAST/EiJ macrophages exhibited earlier cell death than BL6 macrophages following infection (Figure 5F)”

Our strain-specific genomes and annotations significantly improve the accuracy of RNA-seq and single-cell RNA-seq analysis.

“For example, our strain-specific genomes and annotations significantly improve the accuracy of RNA-seq and single-cell RNA-seq analysis.”

There is extensive variation between protein-coding sequences of predicted Apol genes within the chr15 region among the CC/DO founder lines.

“We report extensive variation between protein-coding sequences of predicted Apol genes found within this region among the CC/DO founder lines”

Three distinct scenarios were identified where expression differences occur.

“We identified three distinct scenarios where these differences occur.”

Two quantitative trait loci representative of pulmonary IL-17 levels and bacterial burden at 4 weeks post-infection were mapped to a chr15 locus.

“Two quantitative trait loci (QTLs), representative of pulmonary interleukin (IL)-17 levels and bacterial burden at 4 weeks post-infection, were mapped to a chr15 locus (Figure 5A).”

• AssumptionRests on an assumption: CC strains vary dramatically in their susceptibility to infection and produce qualitatively distinct immune states.

The earlier cell death in CAST/EiJ macrophages corresponded to enhanced Mtb restriction by 4 days post-infection.

“which corresponded to enhanced Mtb restriction by 4 days post-infection (Figure 5G).”

The Irga locus contains several true novel genes in the PWK/PhJ genome.

“Irga locus (blue), in which there are several true novel genes in the PWK/PhJ genome.”

chr15 is highly polymorphic for structural variants compared to the GRCm39 reference genome.

“To measure substantial changes in genomic sequence from the GRCm39 reference genome from BL6, chr15 was scanned for structural variants and was found to be highly polymorphic (Figure 5D).”

Hundreds of non-reference genes discovered and published in historical literature but absent from the BL6-based reference have now been located in these new genomes, including Defa1, Raet1a, and Klra19-28.

“Hundreds of non-reference genes discovered and published in historical literature but absent from the BL6-based reference have now been located in these new genomes, including Defa1, Raet1a, and Klra19-28 (Ly49S, Ly49T, Ly49U, Ly49V, etc.).”

Using the PWK/PhJ reference genome increases the proportion of RNA-seq reads mapped to protein-coding sequence (CDS) for the PWK/PhJ strain compared to GRCm39.

“Increase in proportion of RNA-seq reads mapped to protein-coding sequence (CDS) for the PWK/PhJ strain when using the PWK/PhJ reference genome compared to GRCm39.”

There is a change of pseudogene copy number between GRCm39 and PWK/PhJ at the Fabp5 locus.

“Fabp5 (green), where there is a change of pseudogene copy number between GRCm39 and PWK/PhJ”

Inheritance of the CAST/EiJ haplotype at the chr15 locus was predictive of lower values for both pulmonary IL-17 levels and bacterial burden by BLUP analysis.

“For both phenotypes, inheritance of the CAST/EiJ haplotype at this locus was predictive of lower values by best linear unbiased prediction (BLUP) analysis (Figures 5B and 5C).”

• AssumptionRests on an assumption: CC strains vary dramatically in their susceptibility to infection and produce qualitatively distinct immune states.

This mouse pangenome consists of 17 high-quality reference genomes.

“This mouse pangenome consists of 17 high-quality reference genomes”

We found VNTR-related sequence errors in the CDS of 5 genes in GRCm39.

“We found VNTR-related sequence errors in the CDS of 5 genes in GRCm39.”

In a training and validation set of 262 images, the AI-based VNTR identification tool achieved precision of 93.7% and recall of 83.3%.

“In a training and validation set consisting of 262 images (80% training and 20% validation), precision and recall were found to be 93.7% and 83.3%, respectively.”

Genes cloned or sequenced from non-BL6 mice are often absent from the reference genome.

“Consequently, genes cloned or sequenced from non-BL6 mice are often absent from the reference genome.”

• AssumptionRests on an assumption: Historically, most research was performed using BALB, 129, C3H, DBA, CBA, and FVB lines.

The de novo assembly agrees with the previously published genome structure of all H2 loci.

“Our de novo assembly agrees with the previously published genome structure of all H2 loci (Figures 2A, S2, and S3).”

Genes within the non-syntenic region on chr15 in CAST/EiJ mice meaningfully promote resistance to pathogenic infection.

“Evidence suggests that genes within this region in CAST/EiJ mice meaningfully promote resistance to pathogenic infection.”

• AssumptionRests on an assumption: CC strains vary dramatically in their susceptibility to infection and produce qualitatively distinct immune states.

Currently, our method is not sensitive to short VNTRs because they are not readily visible in the dot plot.

“Currently, our method is not sensitive to short VNTRs because they are not readily visible in the dot plot.”

This coding variation may differentially impact the immunogenetic response to Mtb infection between genetically divergent mouse lines.

“suggesting that this coding variation may differentially impact the immunogenetic response to Mtb infection between genetically divergent mouse lines.”

VNTRs are a relatively unexplored area of mouse genome variation, as they are not easily detected with short-read-based methods.

“VNTRs are a relatively unexplored area of mouse genome variation, as they are not easily detected with short-read-based methods.”

• AssumptionRests on an assumption: Until now, there have been no genome-wide reports of VNTRs in the mouse genome.

We have developed an AI-based method to perform a genome-wide scan of VNTRs in the CDS of all annotated genes among 17 mouse strains.

“We have developed an AI-based method to perform a genome-wide scan of VNTRs in the CDS of all annotated genes among 17 mouse strains.”

• AssumptionRests on an assumption: Until now, there have been no genome-wide reports of VNTRs in the mouse genome.

A double-check of aligned raw data has confirmed multiple low-quality regions in the de novo assembly caused by low PacBio raw read coverage.

“A double-check of aligned raw data has confirmed multiple low-quality regions in the de novo assembly caused by low PacBio raw read coverage.”

Host-driven cell death pathways, regulated by the Apol gene cassette on chr15, promote resistance to pathogenic infection.

“Collectively, this evidence supports previous hypotheses suggesting that host-driven cell death pathways, regulated by the Apol gene cassette on chr15, promote resistance to pathogenic infection.”

• AssumptionRests on an assumption: CC strains vary dramatically in their susceptibility to infection and produce qualitatively distinct immune states.

VNTRs might be a genome phenomenon caused by different reasons and mechanisms.

“In short, VNTRs might be a genome phenomenon caused by different reasons and mechanisms.”