Analyzing a person’s gene expression requires mapping their RNA landscape to a standard reference to gain insight into the degree to which genes are “turned on” and perform functions in the body. But researchers can run into issues when the reference does not provide enough information to allow for accurate mapping, an issue known as reference bias. Researchers at UC Santa Cruz (UCSC) have now reported on what they suggest is the first-ever method for analyzing RNA sequencing data genome-wide using a “pantranscriptome,” which combines a transcriptome and a pangenome—a reference that contains genetic material from a cohort of diverse individuals, rather than just a single linear strand.

The team, headed by UCSC associate professor of Biomolecular Engineering Benedict Paten, PhD, has released a toolkit that allows researchers to map an individual’s RNA data to a much richer reference, addressing reference bias and leading to much more accurate mapping.

“This is pangenome plus transcriptome—that combination has never really been done before until now,” said Jordan Eizenga, PhD, who is co-first author of the researchers’ published paper and a postdoctoral scholar in the UCSC Computational Genomics Lab. “This is the first time anyone has attempted to incorporate the pangenome as a standard feature of the RNA sequencing mapping.” The researchers have made the tools say the tools publicly available—accessed via Github—and say they will aid scientists who are working to understand gene expression through RNA sequencing analysis. “The pantranscriptome consists of a set of haplotype-specific transcripts (HSTs) and is constructed by projecting (lifting over) the transcripts in a transcript annotation onto a set of known haplotypes,” they summarized, in their published report in Nature Methods.

“With this toolkit, we are employing this more diverse data that we can now get from the pangenome to improve the measurement of gene expression data, something that can widely vary between individuals,” Paten said. “The aim is to make the impact of this more diverse data felt on studies that are looking at gene expression, resulting in better analysis for cell models, organoid models, and other research applications.”

Paten, Eizenga and colleagues describe development of the new toolkit in a paper titled “Haplotype-aware transcriptome analyses using spliced pangenome graphs,” in which they stated, “Our bioinformatics pipeline provides a full stack of tools for pantranscriptomic analysis. It can construct pantranscriptomes, map RNA-seq reads to these pantran- scriptomes, and quantify transcription with haplotype resolution.

RNA’s most commonly recognized function is to translate DNA into proteins, but scientists now understand that the vast majority of RNA is noncoding and does not make proteins, but instead can play roles such as influencing cell structure or regulating genes. The entire RNA landscape is known collectively as the transcriptome, and mapping this allows researchers to better understand an individual’s gene expression. “Transcriptome profiling by RNA sequencing (RNA-seq) has matured into a standard and essential tool for investigating cellular state,” the authors noted. “Bio-informatics workflows for processing RNA-seq data generally begin by comparing reads to a reference genome or reference transcriptome. This is an expedient method that makes it practical to analyze the large volume of data produced by high-throughput sequencing.”

The pantranscriptome builds on the emerging concept of “pangenomics” in the genomics field. Typically when evaluating an individual’s genomic data for variation, scientists compare the individual’s genome to that of a reference made up of a single linear strand of DNA bases. Using a pangenome allows researchers to compare an individual’s genome to that of a genetically diverse cohort of reference sequences all at once, sourced from individuals representing a diversity of biogeographic ancestry. This gives the scientists more points of comparison for which to better understand an individual’s genomic variation. “Computational pangenomics has emerged as a powerful methodology for mitigating reference bias,” the team continued. “Pangenomics approaches lean heavily on abundant, publicly available data about common genomic variation for certain species (notably including humans).”

Mapping RNA sequencing data to understand gene expression can be difficult because the RNA sequences are spliced by cellular mechanisms, meaning one set of RNA data can come from non-connected areas of the genome, making it challenging to correctly align them to a reference. These splicing sites are not uniform across the human population, but vary between individuals. It is also difficult to know which haplotype the RNA comes from – whether the group of genes comes specifically from the set of chromosomes inherited from the individual’s mother, or the set inherited from the father.

The new pipeline of open source tools allows researchers to take the spliced segments of an individual’s RNA, map where they align on a pangenome, identify which haplotype the data belongs to, and analyze gene expression.

First, the pipeline identifies which areas of the genome the RNA sequencing data comes from, including the splice sites, and marks those points on the pangenome reference. Those marked points are then compared to a pantranscriptome consisting of haplotype-specific transcripts generated from the reference data contained within the pangenome. This step requires specialized, challenging algorithmic methods.

Finally, it generates estimates of levels of gene expression based on this comparison between the mapped data and the transcripts in the pantranscriptome,  and identifies which haplotypes the genes come from.  “Our toolchain, which consists of additions to the VG toolkit and a standalone tool, RPVG, can construct spliced pangenome graphs, map RNA sequencing data to these graphs, and perform haplotype-aware expression quantification of transcripts  in a pantranscriptome,” the researchers wrote. “First, VG RNA can combine genomic variation data and transcript annotations to construct a spliced pangenome graph. Next, VG MPMAP can align RNA-seq reads to these graphs with high accuracy. Finally, RPVG can use the alignments from VG MPMAP to quantify haplotype-specific transcript expression.”

“It’s definitely a very forward-looking study in that other genome-wide expression methods are not yet really utilizing pangenomes and haplotype information,” said Jonas Sibbesen, PhD, co-first author on the study and a former postdoctoral scholar in the UCSC Computational Genomics Lab who is now an assistant professor at the University of Copenhagen. “We’re now thinking ahead as to what pangenomics might additionally bring to the table in transcriptomic analyses.”

Some downstream applications are already apparent, the team noted in their paper. The  pipeline can be used to study causes of haplotype-specific differential expression. “We demonstrated one such example by investigating genomic imprinting, uncovering suggestive evidence of complex patterns of imprinting at the isoform level,” they noted. The pipeline could be similarly used to study other sources of haplotype-specific differential expression, the scientists further suggested.

Another application is characterizing genotypes and haplotypes in coding regions from RNA-seq data. “We demonstrated this capability by calling genotypes and HLA diplotypes,” they noted, acknowledging that “work is still needed to improve computational efficiency and accuracy in the HLA region.” Nevertheless the investigators stated, “For all of these applications, the VG MPMAP–RPVG pipeline increases the information that is available from RNA-seq data without paired genomic sequencing. This will enable low-cost study designs and deeper reanalyses of existing data.”

Going forward, the researchers are interested in further developing these tools to be useful for downstream informatics analysis, and tailoring the tools for the particularities of research on single-cell data. For now, the group hopes their new toolkit will serve to show how useful using pangenomics-derived analysis can be.

“We need to be able to explain to some researchers how a pangenome reference will benefit them,” Paten said. “This pipeline is really a first go at doing this for RNA, for functional data, for expression data.”