Improved haplotype inference by exploiting long-range linking and allelic imbalance in RNA-seq datasets

BLR: a flexible pipeline for haplotype analysis of multiple linked-read technologies

Linked-read sequencing promises a one-method approach for genome-wide insights including single nucleotide variants (SNVs), structural variants, and haplotyping. We introduce Barcode Linked Reads (BLR), an open-source haplotyping pipeline capable of handling millions of barcodes and data from multiple linked-read technologies including DBS, 10× Genomics, TELL-seq and stLFR. Running BLR on DBS linked-reads yielded megabase-scale phasing with low (<0.2%) switch error rates. Of 13616 protein-coding genes phased in the GIAB benchmark set (v4.2.1), 98.6% matched the BLR phasing. In addition, large structural variants showed concordance with HPRC-HG002 reference assembly calls. Compared to diploid assembly with PacBio HiFi reads, BLR phasing was more continuous when considering switch errors. We further show that integrating long reads at low coverage (∼10×) can improve phasing contiguity and reduce switch errors in tandem repeats. When compared to Long Ranger on 10× Genomics data, BLR showed an increase in phase block N50 with low switch-error rates. For TELL-Seq and stLFR linked reads, BLR generated longer or similar phase block lengths and low switch error rates compared to results presented in the original publications. In conclusion, BLR provides a flexible workflow for comprehensive haplotype analysis of linked reads from multiple platforms.

Interrogating the Human Diplome: Computational Methods, Emerging Applications, and Challenges

Human DNA sequencing protocols have revolutionized human biology, biomedical science, and clinical practice, but still have very important limitations. One limitation is that most protocols do not separate or assemble (i.e., "phase") the nucleotide content of each of the maternally and paternally derived chromosomal homologs making up the 22 autosomal pairs and the chromosomal pair making up the pseudo-autosomal region of the sex chromosomes. This has led to a dearth of studies and a consequent underappreciation of many phenomena of fundamental importance to basic and clinical genomic science. We discuss a few protocols for obtaining phase information as well as their limitations, including those that could be used in tumor phasing settings. We then describe a number of biological and clinical phenomena that require phase information. These include phenomena that require precise knowledge of the nucleotide sequence in a chromosomal segment from germline or somatic cells, such as DNA binding events, and insight into unique cis vs. trans-acting functionally impactful variant combinations-for example, variants implicated in a phenotype governed by compound heterozygosity. In addition, we also comment on the need for reliable and consensus-based diploid-context computational workflows for variant identification as well as the need for laboratory-based functional verification strategies for validating cis vs. trans effects of variant combinations. We also briefly describe available resources, example studies, as well as areas of further research, and ultimately argue that the science behind the study of human diploidy, referred to as "diplomics," which will be enabled by nucleotide-level resolution of phased genomes, is a logical next step in the analysis of human genome biology.

An efficient genotyper and star-allele caller for pharmacogenomics

High-throughput sequencing provides sufficient means for determining genotypes of clinically important pharmacogenes that can be used to tailor medical decisions to individual patients. However, pharmacogene genotyping, also known as star-allele calling, is a challenging problem that requires accurate copy number calling, structural variation identification, variant calling, and phasing within each pharmacogene copy present in the sample. Here we introduce Aldy 4, a fast and efficient tool for genotyping pharmacogenes that uses combinatorial optimization for accurate star-allele calling across different sequencing technologies. Aldy 4 adds support for long reads and uses a novel phasing model and improved copy number and variant calling models. We compare Aldy 4 against the current state-of-the-art star-allele callers on a large and diverse set of samples and genes sequenced by various sequencing technologies, such as whole-genome and targeted Illumina sequencing, barcoded 10x Genomics, and Pacific Biosciences (PacBio) HiFi. We show that Aldy 4 is the most accurate star-allele caller with near-perfect accuracy in all evaluated contexts, and hope that Aldy remains an invaluable tool in the clinical toolbox even with the advent of long-read sequencing technologies.

CYP2C8, CYP2C9, and CYP2C19 Characterization Using Next-Generation Sequencing and Haplotype Analysis: A GeT-RM Collaborative Project

Pharmacogenetic tests typically target selected sequence variants to identify haplotypes that are often defined by star (∗) allele nomenclature. Due to their design, these targeted genotyping assays are unable to detect novel variants that may change the function of the gene product and thereby affect phenotype prediction and patient care. In the current study, 137 DNA samples that were previously characterized by the Genetic Testing Reference Material (GeT-RM) program using a variety of targeted genotyping methods were recharacterized using targeted and whole genome sequencing analysis. Sequence data were analyzed using three genotype calling tools to identify star allele diplotypes for CYP2C8, CYP2C9, and CYP2C19. The genotype calls from next-generation sequencing (NGS) correlated well to those previously reported, except when novel alleles were present in a sample. Six novel alleles and 38 novel suballeles were identified in the three genes due to identification of variants not covered by targeted genotyping assays. In addition, several ambiguous genotype calls from a previous study were resolved using the NGS and/or long-read NGS data. Diplotype calls were mostly consistent between the calling algorithms, although several discrepancies were noted. This study highlights the utility of NGS for pharmacogenetic testing and demonstrates that there are many novel alleles that are yet to be discovered, even in highly characterized genes such as CYP2C9 and CYP2C19.

Differential haplotype expression in class I MHC genes during SARS-CoV-2 infection of human lung cell lines

Introduction

Cell entry of SARS-CoV-2 causes genome-wide disruption of the transcriptional profiles of genes and biological pathways involved in the pathogenesis of COVID-19. Expression allelic imbalance is characterized by a deviation from the Mendelian expected 1:1 expression ratio and is an important source of allele-specific heterogeneity. Expression allelic imbalance can be measured by allele-specific expression analysis (ASE) across heterozygous informative expressed single nucleotide variants (eSNVs). ASE reflects many regulatory biological phenomena that can be assessed by combining genome and transcriptome information. ASE contributes to the interindividual variability associated with the disease. We aim to estimate the transcriptome-wide impact of SARS-CoV-2 infection by analyzing eSNVs.

Methods

We compared ASE profiles in the human lung cell lines Calu-3, A459, and H522 before and after infection with SARS-CoV-2 using RNA-Seq experiments.

Results

We identified 34 differential ASE (DASE) sites in 13 genes (HLA-A, HLA-B, HLA-C, BRD2, EHD2, GFM2, GSPT1, HAVCR1, MAT2A, NQO2, SUPT6H, TNFRSF11A, UMPS), all of which are enriched in protein binding functions and play a role in COVID-19. Most DASE sites were assigned to the MHC class I locus and were predominantly upregulated upon infection. DASE sites in the MHC class I locus also occur in iPSC-derived airway epithelium basal cells infected with SARS-CoV-2. Using an RNA-Seq haplotype reconstruction approach, we found DASE sites and adjacent eSNVs in phase (i.e., predicted on the same DNA strand), demonstrating differential haplotype expression upon infection. We found a bias towards the expression of the HLA alleles with a higher binding affinity to SARS-CoV-2 epitopes.

Discussion

Independent of gene expression compensation, SARS-CoV-2 infection of human lung cell lines induces transcriptional allelic switching at the MHC loci. This suggests a response mechanism to SARS-CoV-2 infection that swaps HLA alleles with poor epitope binding affinity, an expectation supported by publicly available proteome data.

Reconstruction of evolving gene variants and fitness from short sequencing reads

Directed evolution can generate proteins with tailor-made activities. However, full-length genotypes, their frequencies and fitnesses are difficult to measure for evolving gene-length biomolecules using most high-throughput DNA sequencing methods, as short read lengths can lose mutation linkages in haplotypes. Here we present Evoracle, a machine learning method that accurately reconstructs full-length genotypes (R² = 0.94) and fitness using short-read data from directed evolution experiments, with substantial improvements over related methods. We validate Evoracle on phage-assisted continuous evolution (PACE) and phage-assisted non-continuous evolution (PANCE) of adenine base editors and OrthoRep evolution of drug-resistant enzymes. Evoracle retains strong performance (R² = 0.86) on data with complete linkage loss between neighboring nucleotides and large measurement noise, such as pooled Sanger sequencing data (~US$10 per timepoint), and broadens the accessibility of training machine learning models on gene variant fitnesses. Evoracle can also identify high-fitness variants, including low-frequency 'rising stars', well before they are identifiable from consensus mutations.

Minimizer-space de Bruijn graphs: Whole-genome assembly of long reads in minutes on a personal computer

DNA sequencing data continue to progress toward longer reads with increasingly lower sequencing error rates. Here, we define an algorithmic approach, mdBG, that makes use of minimizer-space de Bruijn graphs to enable long-read genome assembly. mdBG achieves orders-of-magnitude improvement in both speed and memory usage over existing methods without compromising accuracy. A human genome is assembled in under 10 min using 8 cores and 10 GB RAM, and 60 Gbp of metagenome reads are assembled in 4 min using 1 GB RAM. In addition, we constructed a minimizer-space de Bruijn graph-based representation of 661,405 bacterial genomes, comprising 16 million nodes and 45 million edges, and successfully search it for anti-microbial resistance (AMR) genes in 12 min. We expect our advances to be essential to sequence analysis, given the rise of long-read sequencing in genomics, metagenomics, and pangenomics. Code for constructing mdBGs is freely available for download at https://github.com/ekimb/rust-mdbg/.

Perspectives on Allele-Specific Expression

Diploidy has profound implications for population genetics and susceptibility to genetic diseases. Although two copies are present for most genes in the human genome, they are not necessarily both active or active at the same level in a given individual. Genomic imprinting, resulting in exclusive or biased expression in favor of the allele of paternal or maternal origin, is now believed to affect hundreds of human genes. A far greater number of genes display unequal expression of gene copies due to cis-acting genetic variants that perturb gene expression. The availability of data generated by RNA sequencing applied to large numbers of individuals and tissue types has generated unprecedented opportunities to assess the contribution of genetic variation to allelic imbalance in gene expression. Here we review the insights gained through the analysis of these data about the extent of the genetic contribution to allelic expression imbalance, the tools and statistical models for gene expression imbalance, and what the results obtained reveal about the contribution of genetic variants that alter gene expression to complex human diseases and phenotypes.