Transcriptomic resources are very much akin to the generation of a reference genome assembly, by providing important baseline functional knowledge for highly relevant tissues of each species. Given the continuous improvement in sequencing technologies, it was not surprising that many papers utilized the latest platforms (e.g., long-read sequencing, single cell RNA-seq) to define RNA transcripts and splice variants, as well as chromatin accessibility and epigenetic modifications at RNA-expressing genes. These efforts reflect the breadth of the community in targeting many farmed species spanning fish, birds and mammals for different tissues, developmental stages, and cell types. For example, RNA-seq-based transcriptomes for 10 tissues or isolated cell populations from chickens was summarized by Overbey et al. Iso-seqbased (Chang et al.) and Nanopore-based (Halstead et al.) transcriptomes of a large number of cattle tissues was generated and used to identify full-length transcripts and alternative splicing isoforms, which were often tissue-specifically expressed. Long-read transcriptome technology was also used by Ali et al. on fourteen tissues to improve the annotation of the Rainbow trout genome, as well as identify splice isoforms associated with traits of economic importance in aquaculture. Similar long-read transcriptome analysis of several tissues in Salmon genome annotation was shown to substantially improve the transcript catalog for this species (Ramberg et al.). A new RNA isoform, circular RNA, was cataloged across public and new RNA-seq datasets for three tissues from sheep, cattle, and pig (Robic et al.), and both tissue-and developmental stage-associated differences in abundance of circular RNAs was detected. Further, RNA-seq analysis of eight specific flow-sorted populations of peripheral blood mononuclear cells (PBMC) was compared to single-cell RNA-seq analysis of PBMC by Herrera-Uribe et al. Both datatypes extended annotation for the pig genome, identified co-expressed genes for all major PBMC types in porcine blood, and showed many specific cell types could be matched to human PBMC cell-specific transcriptomes. Finally, co-expression analysis between RNAs and miRNAs across different stages of spermatogenesis was used to predict miRNA regulatory targets in this important process (de Lima et al.).
Chromatin accessibility patterns and epigenomic modifications were reported as outcomes of the ovine FAANG project. The work from Davenport et al. and Massa et al. is setting high standards for analyzing histone modification, transcription factor binding and/or whole genome-wide methylation analyses. The authors demonstrated that the level of activity at the functional genomic elements found correlated with nearby transcriptomic expression. Further exploration of transcription start sites (Salavati et al.) confirmed the spatial association of active genomic elements and initiation of transcription. Furthermore, E and colleagues used whole genome sequencing and Hi-C to provide mechanistic insights as to the biological basis for polled intersex syndrome (PIS) leads to reproductive disorders in goats.
An ultimate goal of both basic and applied genomics is to connect genotype to phenotype, and multiple groups reported progress in linking genetic variation with the molecular phenotype of RNA expression, which has seen substantial advancement in Genotype-Tissue Expression (GTEx) studies in humans and model species. By analyzing chicken tissues for which both RNA-seq and genomic DNA sequence were available in two populations, Jehl et al. developed thresholds for variant calling and showed the value of existing RNA-seq datasets for reliable SNP detection in allele-specific expression (ASE) and future GTEx studies. In a second report from this group and again investigating chicken RNA-seq data, Degalez et al., reported on the value of haplotypeaware variant annotation and the interest to consider multi-nucleotide variants in the coding regions. Prowse-Wilkins et al. produced and integrated histone modification and CTCF data across six tissues from lactating dairy cows to identify partitions of the genome predicted to comprise functional regions in these tissues. Importantly, they then showed the level of activity of these functional regions were correlated with nearby gene expression and such regions were enriched for putative causal variants. Interestingly, the level of enrichment improved where regions were correlated with the level of expression and was greatest for QTL for milk production traits. This work provided strong evidence for the core hypothesis of the FAANG project; that form follows function and cataloging genome functional elements can be used to find important (e.g. predictive) variation likely causing phenotypic differences.As exemplified by this collection, the efforts produced by groups throughout the world indicate the future of FAANG is very bright. Having said this, the value of the insights provided by the currently more comprehensive efforts in human and biomedical models is clear, and significantly more progress will be needed to fully exploit the public investment in animal agricultural genomics. Especially challenging will be the validation of predicted functional elements and the verification of casual variants associated with complex traits, as each polymorphism may have only a small effect. However, the next major advancements in translation of farm animal genome functional variation into prediction of biological phenotype will come from such precise knowledge of individual genomes.We close by congratulating each of the contributing authors for their outstanding work, and extend our appreciation to all of the reviewers for their time and effort to improve each submission.With sincere gratitude,
