research highlights

GENOMICS One genome, two haplotypes

Two approaches using either fosmid on high-density single-nucleotide polymor- clones or a microfluidic device are used phism arrays. The researchers presented the to tackle the challenge of a haplotype- individual haplotypes of four people, which resolved human genome. allowed them, for example, to directly deter- Although sequencing the six gigabases of mine the human leukocyte antigen haplo- the human genome has become almost triv- types of individuals. ial, sequencing the two times three gigabases One of the main differences between the of the human genome is anything but. High- two methods is the way the DNA is ampli- throughput sequencing technologies, which fied. Shendure’s team used bacteria to amplify allow genome sequencing in a matter of days, Image of Quake and colleagues’ microfluidic the fosmid clones, but Quake and colleagues apply a shotgun approach that can find the device to partition chromosomes. Reprinted from used in vitro multiple displacement amplifica- polymorphisms on homologous chromo- Nature Biotechnology. tion (MDA). Shendure thinks that although somes but cannot determine the haplotypes; in vitro methods are easier to perform, they that is, they cannot tell which polymorphisms chromosomes, they could define 400-kilo- introduce substantial bias in coverage. Quake’s lie together on the same chromosome. base haplotypes, which correlated to team analyzed the bias introduced by MDA Haplotype information is important to 99.9% with HapMap findings. on chromosome 6. They saw hotspots with answer various questions. For example, Two observations stood out for Shendure: strong coverage bias, but Quake does not see whereas a single polymorphism in a gene is one was the ability to “do very basic popula- this as a hindrance to haplotype sequencing. harmless, the combination of two is not; thus tion genetics,” as he called it. The researchers “If you sequence a haploid genome,” he says, one needs to know whether these polymor- saw that haplotypes, not previously seen in “you don’t need the 30–40× coverage you phisms occur on the same allele. HapMap or the 1000 Genomes data, were the need for diploid base calls. We need much less Although several methods for haplo- most enriched for novel variations. “This is coverage if we are calling just one base instead typing single loci exist, there is a dearth something that is highly expected based on of two. We are able to cover as much of the Nature America, Inc. All rights reserved. All rights Inc. America, Nature 1 of methods for haplotyping on a genome- what we know,” says Shendure, “ but seeing chromosome as we want without ridiculous- wide scale. If the genomes of multiple family it empirically was fun.” The other serendipi- ly deep sequencing.” They have not shown

© 201 members are sequenced, one can infer the tous finding was an improvement of de novo sequencing of a whole haploid genome, but haplotype of a closely related individual, or genome assembly. Recent shotgun sequenc- Quake’s prediction is that “it will be easy to one might do population-based inference if ing data of individual genomes have brought sequence genomes in a haploid fashion, bias genotype information on a large cohort is to light sequences that are not present in the notwithstanding.” available. But these methods do not apply reference genome and therefore cannot be Both groups are working on improving to isolated individuals. mapped to a chromosome. Shendure and their approaches. Quake aims to show that Two independent groups have recently colleagues aligned the unmapped reads from the current partitioning and amplification tackled this problem. Jay Shendure and col- their clones to these contigs, then traced them method is indeed sufficient not only for geno- leagues from the University of Washington, back to the fosmid pools they came from and typing but also for sequencing. Shendure is Seattle sequenced haplotype-resolved looked for a shared location in these fosmids. working on an in vitro amplification approach genomes in 400-kilobase increments In several instances this allowed them to that would obviate the need for cloning while (Kitzman et al., 2010), and Stephen Quake anchor previously unmapped contigs. at the same time not introducing the bias of and co-workers from Stanford University Though the resolution obtained with the current MDA methods. genotyped each haploid chromosome from fosmid pool approach is high, the 400-kilo- It is likely that in the near future a hap- a single cell (Fan et al., 2010). base haplotypes cannot be combined into a lotype-resolved genome will be the norm Shendure and colleagues started with whole chromosome. rather than the exception. a fosmid library of genomic DNA of an To achieve the goal of haplotyping an entire Nicole Rusk Indian individual with an insert size of chromosome, Quake and colleagues devel- 40 kilobases and split it into 115 pools oped a microfluidics device that physically Research Papers of ~5,000 clones each so that each pool separates the chromosomes of a single cell in Fan, H. C. et al. Whole-genome molecular haplotyping of single cells. Nat. Biotechnol. 29, 51–57 (2011). covered about 3% of the genome. After metaphase into 48 distinct channels, allow- Kitzman, J.O. et al. Haplotype-resolved genome sequencing the pools, which were derived ing amplification of the DNA in the channels sequencing of a Gujarati Indian individual. to 99% from one of two homologous and retrieval of the material for genotyping Nat. Biotechnol. 29, 59–63 (2011).

nature methods | VOL.8 NO.2 | FEBRUARY 2011 | 107