This is a guest post by Zach Fuller and Molly Przeworski at Columbia University on how they used low-pass sequencing in the preprint ‘Population genetics of the coral Acropora millepora: Towards a genomic predictor of bleaching’.
The symbiotic relationship between corals and photosynthetic algae (family Symbiodiniaceae) underpins the evolutionary success of these reef-building organisms and the diverse marine ecosystems they support. Ecological stress, such as that brought on by increased seawater temperatures, can break down this symbiotic relationship in a process known as “bleaching”. Because the algal symbionts provide the majority of the energy required by the coral host, prolonged periods of bleaching leaves the coral animal starved and can lead to the eventual death of a colony. Over the last several years, mass bleaching events triggered by increased temperatures have decimated coral reef ecosystems worldwide. For example, since 2014 Australia’s Great Barrier Reef (GBR) has lost nearly half of its coral cover as a result of mass bleaching.

Despite recent massive losses, there is substantial phenotypic variation in bleaching response for a number of coral species. In particular, it has been shown that differences in the bleaching response to temperature are at least partially heritable in the coral Acropora millepora — a common species on the GBR and elsewhere in the Indo-Pacific. Like many heritable complex traits in humans and agricultural specie, these differences should thus be predictable from genomic data. To illustrate how population genomics could potentially be applied to coral conservation, we recently used Gencove’s low-pass sequencing analysis to impute the genomes of nearly 200 A. millepora individuals, thereby obtaining genotypes at over 6 million variants. This approach enabled us to cost-effectively perform a genome-wide association study (GWAS) of bleaching response in A. millepora.
The reference haplotype panel used for imputing genotypes is based on high-coverage whole genome sequences of 44 individuals aligned to a chromosome-scale reference genome assembly we recently made freely available. After phasing, these 88 haplotypes make up the imputation reference panel. We then sequenced 193 A. millepora samples to a mean coverage of 1.5x.
To evaluate the accuracy of the Gencove imputation pipeline using this reference panel, 34 samples that we had originally sequenced at high-coverage were included in the low-pass sequencing. We then compared the original high-coverage genotype calls with the imputed genotypes at various genotype probability and minor allele frequency thresholds. Overall, for common variants (minor allele frequency > 0.05) we observed a correlation > 90% for all samples with > 0.5x coverage, for any genotype probability considered. Considering variants with a genotype probability > 0.95, we observed a correlation of ~ 95%, an accuracy similar to what is observed in human data using reference panels of similar size.

We used this genomic data to carry out a GWAS for bleaching response, measured using a standard visual score recorded at the time of collection. No single SNP is genome-wide significant, suggesting that variation in bleaching is not due to common loci of large effect. Nonetheless, the combined estimated effects of variants across the genome are predictive of phenotype: using a cross-validation approach, we demonstrated that a polygenic score constructed from the GWAS results can distinguish the most bleaching tolerant individuals from those that are most susceptible, and can explain > 60% of the total phenotypic variance when combined with other predictors of bleaching, including environmental effects and the dominant symbiont species type.

The imputation pipeline using our reference panel of 88 haplotypes in now publicly available from Gencove, and can be used on any other A. millepora low-pass sequencing data. This pipeline can be used to go directly from sequencing reads in FASTQ files to imputed genotypes at over 6.8 million variants in a VCF file. The resulting genomic data can be used to carry out GWAS for bleaching or other relevant phenotypes for coral conservation in A. millepora.

