Lauren M. McIntyre

Inaccurate allele and gene expression counts due to map bias and genome ambiguity lead to high false positive and false negative rates in studies of allelic imbalance. We demonstrate that long read RNA sequencing (RNA-seq) and straightforward quality control measures can be used to reduce bias in allele counts in case studies from four species: Drosophila melanogaster, a diploid insect; Solanum tuberosum, an autopolyploid plant; Pongo abelii, a highly heterozygous diploid primate, and Homo sapiens. We recommend (i) mapping to a personalized genome to increase the number of allele assignments; (ii) tracking multimapping reads and tuning mapping parameters to ensure accurate allele and gene expression counts; and (iii) evaluating apparent extreme allele bias to identify errors in genome assembly and annotation. We show that these steps can be executed in a straightforward manner and recommend tools for each step.

Summary: How cell type, sex and disease interact and affect gene expression and splicing is an important, but complicated question. Visualizing and testing specific hypotheses around these complex interactions is an important first step to identifying molecular components underpinning complex disease. Using a meta-analytical framework, we develop an analytical path for identifying testable molecular hypotheses of complex interactions between splicing, sex, disease and cell type. We focus on type 1 diabetes (T1D) but the approach is generalizable to any complex disease with defined candidate loci. Previous studies report T1D-associated splicing in candidate genes, differences in disease effects across immune cell types, sex effects on splicing and cell-type-specific splicing. However, identifying and interpreting complex interactions between sex, splicing and disease are challenging. Here we demonstrate how a gene expression study of T1D, designed to evaluate these interactions can be analyzed in a straightforward manner. We find that sex-dependent T1D-associated splicing is markedly more prevalent in CD4⁺ T cells than in CD8⁺ T cells, affecting 72% of T1D candidate genes in CD4⁺ cells compared to 30% in CD8⁺ cells. We pinpoint exons whose rate of inclusion is affected by the interaction of sex and disease. We use long-read RNAseq to identify novel intron retention events and splice sites which are quantified with short-reads leading to a richer description of the regulatory impact of T1D on alternative splicing. We identify a set of candidate isoforms for follow-up molecular studies in BACH2 , a transcription factor known to be relevant in disease prevalence.

Background: Recent advances in long read technologies provide an unprecedented opportunity to study transcript evolution. However, comparative evolutionary studies, even in Drosophila, are limited by inconsistent and incomplete annotation, and the lack of annotated transcript homology. Results: In this study of five species spanning 28 million years ( D. melanogaster , D. simulans , D. yakuba , D. santomea and D. serrata ), we infer transcript homology using reciprocal liftover, and orthology using network analyses, with data validation from long read RNA-seq of male and female head tissue. We build the first genus level annotation, with 15,996 genes and 56,370 transcripts. Expressed transcripts are conserved, 73% of transcript orthologs are detected in all species. Even the improved annotation underestimates the number of genes with alternative transcripts, with 75% of genes expressing multiple structurally diverse transcripts. In a replicated quantitative evaluation of ∼10,000 genes, both male and female-biased transcripts are expressed in 410 ( D. melanogaster ), 608 ( D. simulans ), and 493 ( D. serrata ) genes and in 118 orthologous genes in the D. melanogaster - D. simulans species pair, indicating greater potential for resolution of sexual conflict by alternative transcription than previously appreciated. We identified 605 transcript orthologs conserved for sex bias in the D. melanogaster-D. simulans species pair and of these, 22 male and 19 female-biased transcripts were conserved in sex bias with the outgroup D. serrata, including transcripts of genes involved in brain development, Sxl target Glutamine synthetase 2 and ciboulot . Conclusions: Conserved alternative transcripts suggest that transcriptional diversity is a pervasive driver of the evolution of functional diversity.

From the seminal paper by Meeuwissen et al. 2001 to the GSA Journals Series launch in 2015, the field of Genomic Prediction continues to gain momentum. The field is increasingly dynamic, with new technology increasing the scale and scope of the data available. Significant challenges exist in building computational models. Questions of how to appropriately account for different types of data, and which data improve predictions are interwoven. What is the best path forward? What methods will improve predictions? Authors can submit their rejoinder or start a new discussion on one of the many important topics in the field by submitting a Dialogue and Debate article for peer review at G3.

To understand the relative importance of cis and trans effects on regulation, we crossed multi-parent recombinant-inbred lines (RILs) to a common tester and measured allele-specific gene expression in the offspring. Testing the difference of allelic imbalance between two RIL × Tester crosses is a test of cis or trans, depending on the RIL alleles compared. The study design also enables to separation of two sources of trans variation, genetic and environmental, detected via interactions with cis effects. We demonstrate the effectiveness of this approach in a long-read RNA-seq experiment in female abdominal tissue at two time points in Drosophila melanogaster. Among the 40% of all loci that show evidence of genetic variation in cis, trans effects due to environment are detectable in 31% of loci and trans effects due to genetic background in 19%, with little overlap in sources of trans variation. The genes identified in this study are associated with genes previously reported to exhibit genetic variation in gene expression. Eleven genes in a QTL for thermotolerance, previously shown to differ in expression based on temperature, have evidence for regulation of gene expression regardless of the environment, including the cuticular protein Cpr67B, suggesting a functional role for standing variation in gene expression. This study provides a blueprint for identifying regulatory variation in gene expression, as the tester design maximizes cis variation and enables the efficient assessment of all pairs of RIL alleles relative to the tester, a much smaller study compared to the pairwise direct assessment.

Chromosomal structural changes happen when genomic stability is compromised, such as in disease or in species hybrids. In these contexts, diminished control of repetitive elements has been reported, but the reasons for this are not yet well understood. There are causal associations between repetitive elements and phenotypes such as disease progression, leading us to the hypothesis that chromosomal structure may be affected by transposable elements (TEs). In an intraspecific hybrid Drosophila melanogaster cell line (PnM), the degree of pairing among trans homologous chromosomes was affected by the presence of nearby TEs, in particular, LINE and LTR elements, such as Baggins1 or Gypsy. Chromosomal pairing was significantly lower in windows containing TEs than in windows without any TEs. Pairing was also affected by TEs in mouse, which suggests a possible general association between TEs and pairing that is highly conserved.

Summary: Inaccurate allele and gene expression counts due to map bias and genome ambiguity lead to high false positive and false negative rates in studies of allelic imbalance. We demonstrate that long read RNA-seq and straightforward quality control measures can be used to reduce bias in allele counts in case studies from four species: Drosophila melanogaster , a diploid insect; Solanum tuberosum , an autopolyploid plant; Pongo abelii , a highly heterozygous diploid primate, and Homo sapiens . We recommend 1) mapping to a personalized genome to increase the number of allele assignments 2) tracking multimapping reads and tuning mapping parameters to ensure accurate allele and gene expression counts and 3) evaluating apparent extreme allele bias to identify errors in genome assembly and annotation. We show that these steps can be executed in a straightforward manner and recommend tools for each step.

We propose crossing multi-parent recombinant-inbred-lines (RILs) to a common tester and measuring allele specific gene expression in the offspring. Testing whether allelic imbalance between two RIL x Tester crosses is equal, is a test of cis or trans depending on the RIL alleles compared. The study design also enables to separate two sources of trans variation, genetic and environmental, detected via interactions with cis effects. Examining these components of regulatory variation, we demonstrate this approach in a long-read RNA-seq experiment in female abdominal tissue at two time points in Drosophila melanogaster . Among the 40% of all loci that show evidence of genetic variation in cis, trans effects due to the environment are detectable in 31% of loci and trans effects due to genetic background are detectable in 19% of loci with little overlap in sources of trans variation. The loci identified in this study are associated with loci previously reported to exhibit genetic variation in gene expression in a range of tissues and large population samples, suggesting that there is consistent variation for genetic regulation of gene expression. We show that eleven loci in a QTL for thermotolerance, previously shown to differ in expression based on temperature, have evidence for regulation of gene expression regardless of the environment, including Cpr67B, a cuticular protein suggesting a potential functional role for standing variation in gene expression. This study provides a blueprint for efficiently identifying regulatory variation in gene expression, as the tester design maximizes cis variation and enables the efficient assessment of all pairs of RIL alleles relative to the tester, a much smaller study compared to the pairwise direct assessment.