I found this paper by reading a nice Nature News published one year ago on haplotyping using short linked-reads, and written by the first author of this paper: 'Non-invasive whole genome sequencing of a human fetus' and I chose it because they propose in 2012 a nice protocol, that might be improved a lot with new haplotyping technologies, such as 10x Genomics Chromium.

• Knowledge gap:
• Prenatal test done on maternal plasma usually targets a small range of known disorders, and some of them are invasive.

• Estimation gives around 13% of cell-free DNA in maternal plasma, actually being fetal DNA.

• Strategy:

• In this paper, they propose a non-invasive method to assess fetus whole-genome, getting access to heterozigosity sites, but also de novo mutations.

• On a first trio, they acquired cell-free DNA from maternal plasma, deep-sequencing and haplotype blocks for maternal DNA, and whole-genome for paternal DNA.

• Main findings:

• Good accuracy in finding maternal and paternal heterozigosity sites
• Good sensitivity for de novo detection, but low specificity (10^7 predicted instead of 44). Even after stringent filtering, they manage to increase specificity at the cost of sensitivity.
• Downsampling for measuring impact of sequencing depth of maternal plasma, shorter haplotype blocks, and fetal DNA concentration.
• Validation of their findings on postnatal sample.

--> 4 years later with 10x Genomics library: * works with ~ 100 times less DNA than standard approaches (especially relevant for paternal haplotyping) * get access to better resolution on haplotype blocks (especially the small ones), allowing to assess structural variations.

Discussion: * Interpretability of such a prenatal whole-genome screening. * What kind of decision could be done based on some de novo mutations annotated as pathogen ? * Comparison with a genome screen at adulthood.