Round-up, May 10th – May 22nd

George Church is back in the news, this time for the plan to write a human genome within the next 10 years:

In January, the Department of Health and Human Services updated  what data can be requested by patients about themselves, to cover most things: This was published on a blog and went somewhat unnoticed, but has just come into the limelight as several individuals asked Myriad for all their genetic variants (including benign variants). Myriad initially said No, and although complied after they became aware of the new regulations, the plaintiffs have not dropped their HIPAA complaint. The individuals wanted the benign variants so that they could upload them to ClinVar. Science’s write-up is a must read:

The White House is targeting Mircobiomes for a bold new funding push, with an emphasis on going beyond findings correlations to building tools to get at causation:

The US Equal Employment Opportunity Commission (EEOC) just updated regulations that the ASHG argues means that “employees who decide to keep their own and their spouses’ health information private may have to pay “significantly higher health insurance premiums” than those who share their information”; “”The new EEOC rules mean that Americans could be forced to choose between access to affordable healthcare and keeping their health information private”:

A fascinating read about the relationship between study designers, their participants, the journalists who report on them, and the public reaction:

Natera are finding they are starting to see good reimbursement for average risk pregnancies for their Non-invasive prenatal screening test:

UK start-up PetaGene has a business model based on compressing sequence files, so their customers spend less on storage:

A GenomeWeb interview with the head of BGI Research, focusing on Clinical Tests. One of their projects is a cloud based computer  platform, “we are not just going to offer kits and instruments but also software in the future”:

Happy 10th birthday, Next Generation Sequencing! A review:

A study that looked at concordance between WES and Sanger found the big issue for WES remains poor coverage of certain areas, for which Sanger is still needed — 42 of 51 genes analyzed had “adequate” coverage (over 75% of sequence with coverage of greater than 20×), and for these concordance was very high (97.3%, with not all discrepancies due to WES data):

A study of various labs variant classification using the AMCG/AMP guidelines found high discordance, but that the criteria gave a good framework for talking about differences. The paper does a good job at illustrating how the criteria are being interpreted, including when they are being altered:

A big data-mining project representing joint work from the New York Genomes Center and 23-and-Me has found many variants involved in multiple traits, and some pairs of variants linked causally:

Illumina’s MiSeq does not do well with the sequence ‘CCCGCC’:

A platform to aggregate patient level clinical trail data, Vivli, is gathering steam:

An overview of several studies show human evolution at work: On the shortest of time scales, selection in the middle of the 20th century for an allele that makes it harder to quit smoking has been shown in the UK population.

A method proposed to replace the need to find controls for matched case-control studies, using publicly available data — the concept is termed the Universal Control Repository Network (UNICORN):

Various studies of cancer outcomes from patients who underwent somatic genetic testing have shown less than stellar results, but often on small sample sizes. A meta-analysis of 13,000 patients has shown clear benefit:

The NIH is funding four new studies on the ethical, social, legal implications of genomics:



Round up April 24th – May 9th

Astra-Zeneca are co-ordinating the sequencing of 2m genomes, with multiple partners, including Venter’s HLI, the Sanger, and Genomics England:

Geisinger Health System are sequencing 100,000 exomes (to be expanded to 250,000) – this is an interesting read on a healthcare systems’ genomics ambitions. One advantage they have is the ability to recontact patients to help resolve VUSs:

Foundation Medicine launched a liquid biopsy test, for those cancer patients for whom a tumor sample is unobtainable.

NantHealth is filing for an IPO

The National Human Genome Research Institute will spend $21.8M to assess the cost effectiveness and clinical utility of sequencing

A study that did WGS on 560 breast cancer tumor samples concludes that “the genes harbouring the substantial majority of driver mutations are now known” — there’s about 90 of them. They also identify “genetic signatures” of the cancer– “Each mutational process will leave its own specific pattern on the genome or mutational signature”. A consequence of this is that “statistical models involving mutability cannot assume uniform genomic mutation rates and must consider signature dependent variation as a factor in all future analyses”. BBC write-up: Driver genes:, Signatures:

Memorial Sloan Kettering have announced some of the effects of their MSK-IMPACT tumor-normal panel. They now report on germline variants, and about 20% of patients have pathogenic hereditary cancer variants. Data on the proportion of patients with clinically relevant information returned is not available, except for lung cancer, where it is ~77%.

An overview of hereditary cancer risk, as provided by the American Society of Clinical Oncology for their oncologist members:

Before the ACMG guidelines were published, the International Agency for Research on Cancer (IARC) had a probabilistic framework for interpreting VUSs. A paper that used the two alongside each other found only “slight agreement” between the two methods:

ClinGen have developed a framework for the actionability of disease-gene pairs, and have published the results of applying this framework to the 56 incidental finding genes (see Table 3):

A description from AstraZeneca on how they define actionability in somatic samples that are part of early stage clinical trails:

De novo assembly of a human genome without long reads? 10X and BioNano have joined forces to produce a high quality such assembly based on Ilumina tech:

An easy read on Oxford Nanopore’s technology,, and a more scholarly take on the history of nanopore sequencing:

A Huffington Post blog on the current relevance of genetic testing on mental disorders argues that we are nowhere near ready for prime time: Another blog on personalised medicine in cancer states we are only pursuing this in cancer because ““you are selling hope in a bottle, and you cannot refuse people who are dying”:

The beautiful story of how Thailand have implemented a simple wallet card for the results of a pharmacogenomics test, for which 8% of their population will test positive. The card indicates whether the patient will succumb to an often fatal disease if given a particular drug.

A study that shows that ethnically based Genetic Risk Scores (using aggregates of individual variant effects to predict disease risk) do better than non-ethnically based ones:

A study using ENCODE data to try and elucidate function of intergenic SNPs that are associated with disease: Another study that looks at the combined effects of SNPs on disease risk, focused on Coronary Artery Disease

A review on modeling propensity for common diseases, including environmental factors:

Two alleles associated with having twins have been discovered via GWAS, with large effect sizes:

And finally, how do your genes influence your politics?