Tuesday, March 30, 2021

MGH Study on Genomics, Low-Pass Sequencing, Racial Bias

We periodically hear that too many genetic studies are overpopulated with people of white European backgrounds (e.g. here).   This week, a paper by MGH's Alicia Martin et al. is the basis of a lead story by Christie Rizk at Genomeweb.

  • Genomeweb: Low-Coverage Sequencing Effectively IDs Novel Variants in Underrepresented Populations. (here
  • Amer J Hum Genet: Low-coverage Sequencing Cost-effectively Detects Known and Novel Variation in Underrepresented Populations. (here)
Martin et al. write, "[M]ost genetic studies use genotyping arrays and sequenced reference panels that best capture variation most common in European ancestry populations....Low-coverage sequencing approaches surmount the problems induced by the ascertainment of common genotyping arrays, [and] effectively identify novel variation particularly in underrepresented populations..."

To my knowledge, the largest study directly comparing low pass genome sequencing to microarrays appeard in 2020 in J Molec Diagnostics, by Chaubey et al., a team based at Perkin Elmer (here), using 409 cases, each detected by multiple lab methods.  (See earlier coverage of Chaubey et al., here).   

There is no Category I CPT code yet for low pass cytogenomics, although there are several PLA codes, including one at Mayo Clinic, one at Perkin Elmer, one at New York Genome Center (0012U, 0209U, 0156U.)

More From the Martin et al. Study

Genomeweb goes on to describe the following:
They [Martin et al.] further found that 1x sequencing was among the more affordable options, costing less and performing similarly to or better than commonly used lower-density arrays such as the Illumina GSA. 

They also noted that the GSA is composed of variants that are most common in European populations and so it's therefore not the most appropriate technology for studies of participants with primarily non-European ancestry. 

Aside from cost, low-coverage sequencing had several distinct advantages compared to GWAS arrays, particularly more accurate identification of genetic variation across the allele frequency spectrum in underrepresented populations. 

In the NeuroGAP-Psychosis data, the researchers found that 38 percent of common variants could not be imputed from the 1000 Genomes Phase III data, most likely because of a lack of eastern and southern African diversity in that panel.  


For two Economist articles on racial bias in medical technologies, see here and here.