Co-expression analysis of long non-coding RNAs and mRNAs involved in intramuscular fat deposition in Muchuan black-bone chicken

The intramuscular fat (IMF) content in meat products is positively correlated with meat quality, making it an important consumer trait. Long non-coding RNAs (lncRNAs) play central roles in regulating various biological processes, but little is currently known about the mechanisms by which they regulate IMF deposition in chickens. This study sampled the breast muscles of chickens with high (H) and low (L) IMF content and constructed six small RNA libraries. High-throughput sequencing technology was used to profile the breast muscle transcriptome (lncRNA and mRNA) and to identify the differentially expressed lncRNAs (DELs) and mRNAs (DEGs) between the H and L groups. In total, 263 DELs (118 up-regulated and 145 down-regulated lncRNAs) and 443 DEGs (203 up-regulated and 240 down-regulated genes) were identified between the two groups. To analyse the DELs-DEGs interaction network, co-expression analysis was conducted to identify lncRNA-mRNA pairs. In total, 19,270 lncRNA/mRNA pairs were identified, including 16,398 significant correlation pairs that presented as positive and 2872 pairs that presented as negative. The lncRNA-mRNA network comprised 263 lncRNA nodes and 440 mRNA nodes. Pathway analysis, using the Kyoto Encyclopedia of Genes and Genomes, indicated that pathways associated with fat deposition and lipid metabolism such as the MAPK, PPAR, GnRH, ErbB and calcium signalling pathways, fatty acid elongation and fatty acid metabolism. Overall, the study identified potential candidate lncRNAs, genes and regulatory networks associated with chicken IMF deposition. These findings provide new insights to help clarify the regulatory mechanisms of IMF deposition in chickens which can be used to improve the IMF content in poultry.

An investigation on the influence of various biochemical tenderness factors on eight different bovine muscles

This study's objective was to understand the relationships between biochemical tenderness components and subject/objective tenderness of eight muscles from beef carcasses. Longissimus thoracis (LT), pectoralis profundus (PP), supraspinatus (SS), triceps brachii (TB), gluteus medius (GM), rectus abdominus (RA), rectus femoris (RF), and semitendinosus (ST) were collected from 10 USDA upper 2/3 Choice beef carcasses and assigned to a two- or 21-day aging period (n=160). Troponin-T (TNT) degradation, desmin degradation, sarcomere length, collagen content, mature collagen crosslink density, intramuscular lipid content, pH, Warner-Bratzler Shear Force (WBSF) and trained sensory panel analyses were measured. A Pearson correlation analysis was conducted to determine the relationship between each tenderness contributor measured in this study with WBSF or the overall tenderness evaluated by the trained panelist for each of the eight muscles. In addition, multivariate regression models were constructed to confirm this relationship. The results showed that muscle anatomical locations and physiological functions driven by muscle fiber types may explain some of the biochemical/tenderness differences found in this study. The correlation analysis showed that each muscle had a specific tenderness factor(s) that contributed to the overall tenderness. For instance, tenderness for LT, TB, GM, RA, and ST may be influenced more by proteolytic degradation, while the collagen characteristics may primarily influence tenderness for PP. Also, lipid content has a significant influence on GM tenderness. Finally, the multivariate regression model showed that almost all of the biochemical measurements conducted in this study played a minor but important role as an overall tenderness predictor on a whole carcass basis. Increasing the knowledge base on the various tenderness components' level of contribution will allow end-users to develop specific tenderness management strategies to ensure consistent tenderness in beef products.

Comparison of carcass characteristics and meat quality between Simmental crossbred cattle, cattle-yaks and Xuanhan yellow cattle

BACKGROUND

The study compared the growth performance, carcass characteristics and meat quality of steers of Xuanhan yellow cattle, Simmental crossbreed cattle (Simmental × Xuanhan yellow cattle) and cattle-yak (Jersey × yak). All steers were feed with the same diet from 6 months until slaughter at 30 months. The longissimus dorsi muscle was used to compare the meat quality traits.

RESULTS

By comparison, Simmental crossbreed cattle had higher growth performance (P < 0.05) and carcass characteristics (P < 0.05); cattle-yak had higher value of a*, b* of meat color (P < 0.05) and higher protein contents of meat (P < 0.05); Xuanhan yellow cattle had higher water holding capacity (P < 0.05) and lower shear force (P < 0.05).

CONCLUSIONS

The results show that Simmental crossbred cattle had better meat performance and provided low-fat meat with a beneficial fatty acid composition, but with lower meat quality; cattle-yaks provided greater meat color and higher protein content; Xuanhan yellow cattle provided meat with preferable tenderness. © 2020 Society of Chemical Industry.

Comparison of Growth Performance and Meat Quality Traits of Commercial Cross-Bred Pigs versus the Large Black Pig Breed

The meat quality of different pig breeds is associated with their different muscle tissue physiological processes, which involves a large variety of genes related with muscle fat and energy metabolism. Understanding the differences of biological processes of muscle after slaughter is helpful to reveal the meat quality development of different breeds. Therefore, eight native Large Black pigs (BP), with high fat content in meat, and seven cross-bred commercial pigs (CP), which had a high feed efficiency with high lean meat, were used to investigate the differences in their meat quality and RNA transcriptomes. The average daily gain (ADG) and hot carcass weight (HCW) of CP were higher than BP, but the back-fat thickness of BP was higher than CP (p < 0.05). The CP had higher a* (redness) but lower h (hue angle) than BP (p < 0.05). The metmyoglobin (MMb) percentage of CP was higher (p < 0.05) than BP. The fat content and oxygen consumption of longissimus dorsi (LD) muscles in BP were higher (p < 0.05) than CP. BP had higher monounsaturated fatty acids (MUFA) content, but CP had higher polyunsaturated fatty acids (PUFA) content (p < 0.05). The RNA-seq data highlighted 201 genes differentially expressed between the two groups (corrected false discovery rate (FDR) p < 0.05), with 75 up-regulated and 126 down-regulated genes in BP compared with CP using the fold change (FC). The real-time PCR was used to validate the results of RNA-seq for eight genes, and the genes related to lipid and energy metabolism were highly expressed in BP (p < 0.05). Based on the results, BP had superior intramuscular fat content to CP, while the growth performance of CP was better, and the transcriptomic differences between these two groups of pigs may cause the meat quality and growth performance variance.

A Meta-Analysis Including Pre-selected Sequence Variants Associated With Seven Traits in Three French Dairy Cattle Populations

A within-breed genome-wide association study (GWAS) is useful when identifying the QTL that segregates in a breed. However, an across-breed meta-analysis can be used to increase the power of identification and precise localization of QTL that segregate in multiple breeds. Precise localization will allow including QTL information from other breeds in genomic prediction due to the persistence of the linkage phase between the causal variant and the marker. This study aimed to identify and confirm QTL detected in within-breed GWAS through a meta-analysis in three French dairy cattle breeds. A set of sequence variants selected based on their functional annotations were imputed into 50 k genotypes for 46,732 Holstein, 20,096 Montbeliarde, and 11,944 Normande cows to identify QTL for milk production, the success rate at insemination of cows (fertility) and stature. We conducted within-breed GWAS followed by across-breed meta-analysis using a weighted Z-scores model on the GWAS summary data (i.e., P-values, effect direction, and sample size). After Bonferroni correction, the GWAS result identified 21,956 significantly associated SNP (P _FWER < 0.05), while meta-analysis result identified 9,604 significant SNP (P _FWER < 0.05) associated with the phenotypes. The meta-analysis identified 36 QTL for milk yield, 48 QTL for fat yield and percentage, 29 QTL for protein yield and percentage, 13 QTL for fertility, and 16 QTL for stature. Some of these QTL were not significant in the within-breed GWAS. Some previously identified causal variants were confirmed, e.g., BTA14:1802265 (fat percentage, P = 1.5 × 10^-760; protein percentage, P = 7.61 × 10^-348) both mapping the DGAT1-K232A mutation and BTA14:25006125 (P = 8.58 × 10^-140) mapping PLAG1 gene was confirmed for stature in Montbeliarde. New QTL lead SNP shared between breeds included the intronic variant rs109205829 (NFIB gene), and the intergenic variant rs41592357 (1.38 Mb upstream of the CNTN6 gene and 0.65 Mb downstream of the CNTN4 gene). Rs110425867 (ZFAT gene) was the top variant associated with fertility, and new QTL lead SNP included rs109483390 (0.1 Mb upstream of the TNFAIP3 gene and 0.07 Mb downstream of PERP gene), and rs42412333 (0.45 Mb downstream of the RPL10L gene). An across-breed meta-analysis had greater power to detect QTL as opposed to a within breed GWAS. The QTL detected here can be incorporated in routine genomic predictions.

Multibreed genome wide association can improve precision of mapping causative variants underlying milk production in dairy cattle

BACKGROUND

Genome wide association studies (GWAS) in most cattle breeds result in large genomic intervals of significant associations making it difficult to identify causal mutations. This is due to the extensive, low-level linkage disequilibrium within a cattle breed. As there is less linkage disequilibrium across breeds, multibreed GWAS may improve precision of causal variant mapping. Here we test this hypothesis in a Holstein and Jersey cattle data set with 17,925 individuals with records for production and functional traits and 632,003 SNP markers.

RESULTS

By using a cross validation strategy within the Holstein and Jersey data sets, we were able to identify and confirm a large number of QTL. As expected, the precision of mapping these QTL within the breeds was limited. In the multibreed analysis, we found that many loci were not segregating in both breeds. This was partly an artefact of power of the experiments, with the number of QTL shared between the breeds generally increasing with trait heritability. False discovery rates suggest that the multibreed analysis was less powerful than between breed analyses, in terms of how much genetic variance was explained by the detected QTL. However, the multibreed analysis could more accurately pinpoint the location of the well-described mutations affecting milk production such as DGAT1. Further, the significant SNP in the multibreed analysis were significantly enriched in genes regions, to a considerably greater extent than was observed in the single breed analyses. In addition, we have refined QTL on BTA5 and BTA19 to very small intervals and identified a small number of potential candidate genes in these, as well as in a number of other regions.

CONCLUSION

Where QTL are segregating across breed, multibreed GWAS can refine these to reasonably small genomic intervals. However, such QTL appear to represent only a fraction of the genetic variation. Our results suggest a significant proportion of QTL affecting milk production segregate within rather than across breeds, at least for Holstein and Jersey cattle.

Fatty acid profiles and adipogenic gene expression of various fat depots in Japanese Black and Holstein steers

Objective of the study was to assess the breed effect on fatty acid (FA) composition of different adipose tissues and on mRNA expression of genes involved in adipogenesis and fat metabolism. Japanese Black (JB) and Holstein (HS) steers were kept under equivalent conditions with high energy intake resulting in large differences in intramuscular fat (IMF) accumulation in longissimus muscle (LM). The relative FA composition of muscle, intermuscular fat, visceral fat, and perirenal fat was comparable between JB and HS steers. Circulating fatty acids were also similar in both breeds. Most relevant breed effects were identified in IMF, underlining the uniqueness of this adipose tissue site. JB steers had more monounsaturated FA and less saturated FA. Perilipin 1 and adipose differentiation-related protein (ADFP) mRNA levels were higher in IMF of JB. The results suggest advanced maturity of IMF cells in JB and altered local conditions in muscle influencing IMF accumulation and composition.

The GENOTEND chip: a new tool to analyse gene expression in muscles of beef cattle for beef quality prediction

Background

Previous research programmes have described muscle biochemical traits and gene expression levels associated with beef tenderness. One of our results concerning the DNAJA1 gene (an Hsp40) was patented. This study aims to confirm the relationships previously identified between two gene families (heat shock proteins and energy metabolism) and beef quality.

Results

We developed an Agilent chip with specific probes for bovine muscular genes. More than 3000 genes involved in muscle biology or meat quality were selected from genetic, proteomic or transcriptomic studies, or from scientific publications. As far as possible, several probes were used for each gene (e.g. 17 probes for DNAJA1). RNA from Longissimus thoracis muscle samples was hybridised on the chips. Muscles samples were from four groups of Charolais cattle: two groups of young bulls and two groups of steers slaughtered in two different years. Principal component analysis, simple correlation of gene expression levels with tenderness scores, and then multiple regression analysis provided the means to detect the genes within two families (heat shock proteins and energy metabolism) which were the most associated with beef tenderness. For the 25 Charolais young bulls slaughtered in year 1, expression levels of DNAJA1 and other genes of the HSP family were related to the initial or overall beef tenderness. Similarly, expression levels of genes involved in fat or energy metabolism were related with the initial or overall beef tenderness but in the year 1 and year 2 groups of young bulls only. Generally, the genes individually correlated with tenderness are not consistent across genders and years indicating the strong influence of rearing conditions on muscle characteristics related to beef quality. However, a group of HSP genes, which explained about 40% of the variability in tenderness in the group of 25 young bulls slaughtered in year 1 (considered as the reference group), was validated in the groups of 30 Charolais young bulls slaughtered in year 2, and in the 21 Charolais steers slaughtered in year 1, but not in the group of 19 steers slaughtered in year 2 which differ from the reference group by two factors (gender and year). When the first three groups of animals were analysed together, this subset of genes explained a 4-fold higher proportion of the variability in tenderness than muscle biochemical traits.

Conclusion

This study underlined the relevance of the GENOTEND chip to identify markers of beef quality, mainly by confirming previous results and by detecting other genes of the heat shock family as potential markers of beef quality. However, it was not always possible to extrapolate the relevance of these markers to all animal groups which differ by several factors (such as gender or environmental conditions of production) from the initial population of reference in which these markers were identified.

Identification and utilization of genes associated with beef qualities

The fatty acid composition of adipose tissue in beef has been recognized as an important trait because of its relationship with beef quality, including favorable beef flavor and tenderness. Over the last decade, we have tried to identify the genes responsible for the fatty acid composition in cattle, and have found the following. (i) Genetic polymorphism of stearoyl-CoA desaturase (SCD) is one of the responsible genes associated with fatty acid composition. The average effects of gene substitution of the SCD type A gene on the monounsaturated fatty acid (MUFA) percentage and the melting point of intramuscular fat were approximately +1.0% and -1.0°C, respectively. (ii) Intron polymorphism of sterol regulatory element binding protein-1 (SREBP-1) also affected MUFA. (iii) No effect of SCD or SREBP-1 genotypes on any representative carcass traits of Japanese Black in the field population was observed. (iv) Additional genetic markers adipocytes fatty acid binding protein 4 (FABP4) and liver X receptor α also affected the fatty acid composition. (v) SCD and FABP4 significantly affected fatty acid composition in Holstein steers. These findings will bring new insight into the fat-related carcass traits of beef cattle and will thus contribute to the beef industry.

Use of a bovine genome array to identify new biological pathways for beef marbling in Hanwoo (Korean Cattle)

BACKGROUND

Marbling (intramuscular fat) is a valuable trait that impacts on meat quality and an important factor determining price of beef in the Korean beef market. Animals that are destined for this high marbling market are fed a high concentrate ration for approximately 30 months in the Korean finishing farms. However, this feeding strategy leads to inefficiencies and excessive fat production. This study aimed to identify candidate genes and pathways associated with intramuscular fat deposition on highly divergent marbling phenotypes in adult Hanwoo cattle.

RESULTS

Bovine genome array analysis was conducted to detect differentially expressed genes (DEGs) in m. longissimus with divergent marbling phenotype (marbling score 2 to 7). Three data-processing methods (MAS5.0, GCRMA and RMA) were used to test for differential expression (DE). Statistical analysis identified 21 significant transcripts from at least two data-processing methods (P < 0.01). All 21 differentially expressed genes were validated by real-time PCR. Results showed a high concordance in the gene expression fold change between the microarrays and the real time PCR data. Gene Ontology (GO) and pathway analysis demonstrated that some genes (ADAMTS4, CYP51A and SQLE) over expressed in high marbled animals are involved in a protein catabolic process and a cholesterol biosynthesis process. In addition, pathway analysis also revealed that ADAMTS4 is activated by three regulators (IL-17A, TNFα and TGFβ1). QRT-PCR was used to investigate gene expression of these regulators in muscle with divergent intramuscular fat contents. The results demonstrate that ADAMTS4 and TGFβ1 are associated with increasing marbling fat. An ADAMTS4/TGFβ1 pathway seems to be associated with the phenotypic differences between high and low marbled groups.

CONCLUSIONS

Marbling differences are possibly a function of complex signaling pathway interactions between muscle and fat. These results suggest that ADAMTS4, which is involved in connective tissue degradation, could play a role in an important biological pathway for building up marbling in cattle. Moreover, ADAMTS4 and TGFβ1could potentially be used as an early biological marker for marbling fat content in the early stages of growth.

Hyaluronan: is it a biomarker for adipose development within bovine muscle?

Based on an association with extracellular matrix remodelling, mitosis, proliferation and adipogenic differentiation, the glycosaminoglycan hyaluronan (HA) was assessed as a marker for intramuscular fat (IMF) development (marbling) in bovine loin muscle (longissimus dorsi, LD). Loin samples collected from the quartering site of feedlot-finished Wagyu–Angus and Jersey–Limousin steers were assayed for percentage IMF (IMF%) and HA after assignment of AUS-MEAT marbling scores. There was a moderate phenotypic correlation (r2 = 0.69) between IMF% and marbling score but little variance was explained by HA concentration. Breed was not a significant factor in marbling score or IMF% but did influence the HA concentration of the LD, with Wagyu–Angus steers having 2-fold more HA than Jersey–Limousin steers at the same marbling score. The non-linear decline in fat-adjusted HA levels as marbling score increased suggests that HA concentration was associated with lean growth potential of the muscle rather than adipogenesis. Using a different experimental approach, differences in distribution and amount of HA could not be discerned in histological sections of LD from age-matched Wagyu–Hereford heifers allocated to a low (score 1) or medium (score 3) marbling score group. These findings were consistent with the absence of differences between the two groups for other indicators of fatness (IMF% and P8 fat depth), maturity and myofibre characteristics despite an increase in oxidative capacity of the muscle with age. The data support the conclusion that the concentration of HA in the LD alone was not predictive of development of intramuscular fat.

Differentially-expressed genes in pig Longissimus muscles with contrasting levels of fat, as identified by combined transcriptomic, reverse transcription PCR, and proteomic analyses

Intramuscular fat content is important for many meat quality parameters. This work is aimed at identifying functional categories of genes associated with natural variation among individuals in intramuscular fat content to help the design of genetic schemes for high marbling potential. Taking advantage of the global nature of transcriptomic and proteomic technologies, 40 genes were identified as differently expressed between high fat and low fat pig Longissimus muscles at slaughter weight. They are involved in metabolic processes, cell communication, binding, and response to stimulus. Using real-time PCR in muscle biopsies taken earlier in the fattening period, the group with a high intramuscular fat content was also characterized by the down-expression of genes playing a negative role in adipogenesis, such as architectural transcription factor high-motility hook A1, mitogen activated protein-kinase14, and cyclin D1. These results suggest that interindividual variability in intramuscular fat content might arise essentially from differences in early adipogenesis.

Application of gene expression studies in livestock production systems: a European perspective

In the context of sustainable agriculture and animal husbandry, understanding animal physiology remains a major challenge in the breeding and production of livestock, especially to develop animal farming systems that respond to the new and diversified consumer demand. Physiological processes depend on the expression of many genes acting in concert. Considerable effort has been expended in recent years on examining the mechanisms controlling gene expression and their regulation by biological and external factors (e.g. genetic determinants, nutritional factors, and animal management). Two main strategies have been developed to identify important genes. The first one has focussed on the expression of candidate genes for key physiological pathways at the level of both the transcripts and proteins. An original strategy has emerged with the advent of genomics that addresses the same issues through the examination of the molecular signatures of all genes and proteins using high-throughput techniques (e.g. transcriptomics and proteomics). In this review, the application of the gene expression studies in livestock production systems is discussed. Some practical examples of genomics applied to livestock production systems (e.g. to optimise animal nutrition, meat quality or animal management) are presented, and their outcomes are considered. In the future, integration of the knowledge gained from these studies will finally result in optimising livestock production systems through detection of desirable animals and their integration into accurate breeding programs or innovative management systems.

Adipogenesis of bovine perimuscular preadipocytes

In this study, non-transformed progeny adipofibroblasts, derived from mature adipocyte dedifferentiation, was used as a novel in vitro model to study adipogenic gene expression in cattle. Adipofibroblasts from dedifferentiated mature perimuscular fat (PMF) tissue were cultured with differentiation stimulants until the cells exhibited morphological differentiation. Treated cells were harvested from day 2 to 16 for RNA extraction, whereas control cells were cultured without addition of stimulants. Results from time course gene expression assays by quantitative real-time PCR revealed that peroxisome proliferator-activated receptor gamma (PPAR-gamma), sterol regulatory element binding protein 1 (SREBP-1) and their six down-stream genes were co-expressed at day 2 post-differentiation induction. When compared to other adipogenesis culture systems, the adipogenic gene expression of bovine PMF adipofibroblasts culture was different, especially to the rodent model. Collectively, these results demonstrated PPAR-gamma and SREBP-1 cooperatively play a key role to regulate the re-differentiation of bovine adipofibroblasts, during early conversion stages in vitro.

Gene expression profiling of Hereford Shorthorn cattle following challenge with Boophilus microplus tick larvae

The ability of cattle to resist tick infestations is partly genetically determined. In order to better define the nature of Bos taurus resistance to the cattle tick Boophilus microplus, skin gene expression was studied using a cattle skin derived cDNA microarray. Expression profiles were determined in skin biopsies sampled from three highly tick resistant animals (HR) and two animals with lower tick resistance (LR) at time 0, immediately before challenge, and again 24 h after challenge. The analysis of the resulting expression data addressed two biological questions: first, for any animal exposed to ticks, which genes are differentially expressed in the 24 h following challenge; and second, which genes are differentially expressed between animals of high and low resistance at 24 h after challenge? In total, 214 genes were found to be differentially expressed in response to larval challenge across all the animals. Seventy-two genes were upregulated and 76 were downregulated at 24 h after challenge. Genes with significantly altered gene expression levels following tick infestation were predominantly keratin genes or mitochondrial genes, as well as odorant binding protein (OBP) and Bos taurus major allergen BDA20. In addition, we identified 66 genes with differential expression between HR and LR animals at 24 h. Of these, genes representing the extracellular matrix and immunoglobulin gene expression pathways were overrepresented. Three differentially expressed genes, OBP, Bos taurus major allergen BDA20 and dendritic cell protein HFL-B5 were further analysed by quantitative reverse transcription PCR (qRT-PCR). The qRT-PCR assay results closely mirrored the expression profiles found in the microarray experiment.

A gene coexpression network for bovine skeletal muscle inferred from microarray data

We present the application of large-scale multivariate mixed-model equations to the joint analysis of nine gene expression experiments in beef cattle muscle and fat tissues with a total of 147 hybridizations, and we explore 47 experimental conditions or treatments. Using a correlation-based method, we constructed a gene network for 822 genes. Modules of muscle structural proteins and enzymes, extracellular matrix, fat metabolism, and protein synthesis were clearly evident. Detailed analysis of the network identified groupings of proteins on the basis of physical association. For example, expression of three components of the z-disk, MYOZ1, TCAP, and PDLIM3, was significantly correlated. In contrast, expression of these z-disk proteins was not highly correlated with the expression of a cluster of thick (myosins) and thin (actin and tropomyosins) filament proteins or of titin, the third major filament system. However, expression of titin was itself not significantly correlated with the cluster of thick and thin filament proteins and enzymes. Correlation in expression of many fast-twitch muscle structural proteins and enzymes was observed, but slow-twitch-specific proteins were not correlated with the fast-twitch proteins or with each other. In addition, a number of significant associations between genes and transcription factors were also identified. Our results not only recapitulate the known biology of muscle but have also started to reveal some of the underlying associations between and within the structural components of skeletal muscle.

Gene expression-based approaches to beef quality research

Advances in mammalian genomics have permitted the application of gene expression profiling approaches to gene discovery for meat quality traits in cattle. The first custom cDNA microarray based on the transcriptome of bovine muscle and fat tissue was developed and applied to animal experimentation and cell culture experimentation between 1999 and 2005. Complementary DNA microarray tools for beef quality research were developed in parallel with bioinformatics tools that permit the analysis of microarray data obtained from complex experimental designs commonly encountered in large animal research. In addition, tools were designed to link gene expression data with gene function in the bovine, such as in vitro models of bovine adipogenesis and bioinformatics tools to map gene networks from expression data. The application of these genomics tools to the study of beef quality has yielded novel knowledge of genes and molecules involved in the processes of intramuscular adipogenesis and protein turnover. This review summarises the current state of knowledge and important lessons derived from bovine genomics initiatives in Australia and around the world.

Construction of gene interaction and regulatory networks in bovine skeletal muscle from expression data

We propose a data-driven reverse engineering approach to isolate the components of a gene interaction and regulatory network. We apply this method to the construction of a network for bovine skeletal muscle. Key nodes in the network include muscle-specific genes and transcription factors. muscle-specific genes are identified from data mining the USA National Cancer Institute, Cancer Genome Anatomy Project database, while transcription factors are predicted by accurate function annotation. A total of 5 microarray studies spanning 78 hybridisations and 23 different experimental conditions provided raw expression data. A recently-reported analytical method based on multivariate mixed-model equations is used to compute gene co-expression measures across 624 genes. The resulting network included 102 genes (of which 40 were muscle-specific genes and 7 were transcription factors) that clustered in 7 distinct modules with clear biological interpretation.