Round-up May 31st – July 9th

I have slipped in my resolution for a weekly round-up. Here is another catch-up. And there has certainly been a lot to catch up on!

Germline news

In the first ever randomized controlled trial of whole genome sequencing for healthy adults, which reported on variation across genes linked to Mendelian disease, many participants were found to have variants believed to definitively cause disease by adulthood, even though they were living disease free. STAT has a write-up.

At one point, it was thought that diseases such as obesity, or traits such as height might be polygenic in nature, i.e. affected by variation in some set of closely involved genes. Genome Wide Association Studies are premised on this idea. But it is challenged by a recent paper that puts forward the case for the omnigenic model i.e. almost every gene effects every trait, because of how interconnected the networks are. If true, this suggests that we should be concentrating on mapping out the networks operating in different cell types. Ed Yong has a good write-up.

Veritas has launched myBabyGenome in China. For $1500 parents can find out about a series of traits (e.g. how novelty seeking” their child may be) and disease risks. But as genetics prof Jim Evans says You run the risk of predestination based on bad science”. They won’t be told about e.g. variation that leads to an increased risk if Alzheimers. Instead, they will be offered the chance to purchase that information later on.

A study of families in Scotland involving 20,000 individuals reports that brainier people have fewer variants that impair general health and intelligence, rather than possessing variants that make them smarter. While twin studies suggest that 50-80% of intelligence is based on genetics, studies to date had been able to attribute only about 30% of the variance to genetics the gap is known as the mystery of missing heritability”. This study was able to explain this gap, because it could isolate the effect of very rare variants usually too rare to be missed, but because of their presence in families, detectable by this study. The results are support for the mutational load” paradigm for thinking of genetic endowment.

Huntington’s sufferers have one normal and one altered copy of the gene HTT. A study that used CRISPR-Cas9 to knock out both copies of HTT in mice found no negative effects and improvements in symptoms.

Over 1000 new reference genomes of bacteria and archaea have been deposited by the JGI.

New York fertility doctor, Dr Zhang, who evaded US regulations and went to Mexico to oversee use of the three-person baby technique, has started a company to market this technique. It will market not only to couples wishing to avoid passing on mitochondrial disease, but also for age-related issues.

A study that looked at the combination of pre-implantation genetic diagnosis (PGD, performed for Mendelian diseases such as Cystic Fibrosis) and screening (PGS, performed to test for aneuploidies such as Downs), found that the combination resulted in more pregnancies given embryo transfers, but fewer transfers, as for more couples no viable embryos remained. After PGD, 56% of embryos were deemed transferable. After PGS, only 27.5% of blastocysts were deemed transferable. Standardly, about 34% of transfers result in pregnancies. Following both PGD and PGS in this study, the number was 49%.

How much of heritability comes from common (>5% frequency) variants? A new study suggests about 43%.

Why hasn’t natural selection reduced the amount of coronary heart disease? This study provides novel evidence that CAD has been maintained in modern humans as a by-product of the fitness advantages those genes provide early in human lifecycles” paper here

Analysis of mummy DNA suggests Ancient Egyptians individuals were more similar to present day people from the Near East than to populations residing in Egypt today

The Precision Medicine Initiative, launched in 2015, has just started beta enrollment in its All Of Us program, which aims to be a 1 million strong cohort. 

Age-related somatic mutations have typically been of interest because of cancer. But they are also of potential relevance for other disease types. A large-scale study finds that a particular type of somatic change in blood cells doubles the risk of coronary heart disease. Which is a good segway into…


Somatic related news

Encouraging results from a small cancer-vaccine study: 6 melanoma patients given a cancer vacine unique to their tumor saw no recurrence. The technique works by first sequencing the tumor, to work out which sections of the DNA make antigens that the immune system is most likely to respond to, and that are different from germline cells. Next, millions of copies of each of 13-20 of these neoantigen regions are injected.  The idea is that the patient’s immune system learns to recognize these sequences as other”, and will then target the cancerous cells.

Evidence that PARP inhibitors perform better than chemotherapy for women with metastatic breast cancer who also have germline BRCA mutations. There are three PARP inhibitors approved for ovarian cancer on the market.

A study of a regulatory region that is well conserved between humans and mice finds when this region is deleted a) mice develop normally, and b) mice are resistant to tumor development, making it a promising drug target. 

Following Keytruda’s approval based on a bimarker indication across all cancer types, new FDA commissioner Gottleib has promised more guidance to allow more drugs to follow this path. 

Flatiron and Foundation have shown the utility of their joint molecular-clinical dataset, including the demonstration that those with an actionable variant have a survival time of 35 months, compared to 19 months for those without.

 A nice synopsis of oncologists’ thinking about genetic testing, arising out of ASCO.

Merck paused two studies of Keytruda in combination with other therapies for multiple myeloma, because of unexplained deaths.

Accurate software for distinguishing germline variants from a tumor without the need for a matched normal sample.


Round-up May 22nd-30th

A study published today showing CRISPR off-target effects knocked CRISPR stocks, despite the fact that the findings were old news. 

The FDA have approved the first therapy based on molecular feature rather than tissue of origin. About 4% of all advanced cancers have the genetic characteristics involved. Merck won the apporval based on an accelerated” process, off the back of a trial of 149 patients.

The Broad have released the latest version of their genome analysis software, GATK, under an open source license. The software, which has 45,000 users, was previously costly for non-academic use.

A meta-analysis of the genetics behind intelligence identifies several genetic loci that collectively explain 4.8% of the variance in intelligence. An editorial in Nature argues that we have historical reasons to fear the field of intelligence studies (history of racism, history of eugenics, fears of biological determinism), but we shouldn’t this hold us back from showing that there is no genetic basis for discrimination”.

When it comes to complex disease, GWAS have historically found relatively common variants of small effect. Larger sample sizes are a way to go, as is using isolated populations in the hope of identifying founder mutations, as this study does to uncover a variant protective against heart disease.

How rare is rare enough for a variant to be possibly causative of disease? This all important cut-off affects how many variants must be considered when evaluating a case. The paper from the ExAC group that lays out a framework for defining this cut-off is finally out, though of course it was up on BioRxiv months ago, and was previously rounded-up here.

Nature reports that police in Xinjiang, China, have been collecting blood samples and are building up DNA sequencing facilities that go beyond what would be needed for regular forensics. There is no clear framework for legal use of the data.

NPR reports on CRISPR use outside the lab. 

Catch-up! April 4th – May 21st

I’ve moved jobs and had a month traveling since my last round-up. An attempt to capture the most significant happenings over the intervening few weeks below, before resuming my regular weekly schedule.
I am now working for Driver, with a mission to cure cancer, so expect to see an increased emphasis on all things somatic.
Having been blocked from reporting health results for a few years, 23andMe are now allowed to report on 10 diseases direct to consumer, with no physician in the loop. The list does not include tests whose results cold affect treatment decisions (e.g. BRCA1/2)
  • the ACMG are opposed
  • those that they’re at higher risk for Alzheimers are over 6 times as likely to purchase long-term care insurance. Individuals are supposed to share such information with their potential insurer (unlike for health insurance), but the industry fears people won’t (New York Times).
  • on the average, individuals do have measurable lifestyle improvements, and 25% credit their genetic results (improvements were not associated with any particular genetic result). This is in contrast to previous studies, which found no improvement
More than two thirds of ~100 oncologists said they would prescribe immunotherapy rather than a targeted molecular therapy, because the shot of a durable response from the former outweighs the fact that resistant genotypes will be positively selected for in the latter. This follows on from a survey reporting 70% of 132 oncologists said genomic testing was below expectations’.
Two studies on 100 non small-cell lung cancer (NSCLC) cancer patients (the TRACERx study):
  • Samples from many regions finds elevated CNV heterogeneity associated with shorter survival (NEJM
  • Using the tumor sequencing results, the team designed an individualized multiplex-PCR test for each patient, and use the results to show they can predict patient responses/outcomes in some cases (Nature). They also find predictors of which patients have more circulating tumor DNA.
MSK have reported that their IMPACT test for cancer patients identifies clinically relevant mutations for 37% of patients. 11% of patients enrolled on a clinical trial as a result of a variant found. They are optimistic about predicting response to immunotherapy boosting these numbers further (Study N=5009, currently about 17,000 patients).
A study of 100,000 cancers has found that tumor mutational burden (TMB) can be assessed accurately from just ~1.1 Mb of sequence, and report recurrent promoter mutations in PMS2 associated to high TMB.
A study in Nature of the whole genomes of different types of melanoma reveals diverse carcinogenic processes across its subtypes, some unrelated to sun exposure, and reports that most cancers had actionable mutations.
Noninvasive prenatal screening (NIPS) started with chromosomal abnormalities (now reccomended for all pregnancies), and has more recently been extended to the smaller scale microdeletions. Now Natera say they are developing a noninvasive prenatal screening test that will cover  de novo mutations in 30 genes, which will cover diseases that collectively cover diseases found in 1 in 600 births. 
Broad researchers introduced Sherlock via a Science paper: it uses Cas13a, a CRISPR associated protein that targets RNA rather than DNA, as part of a very sensitive molecular diagnostic to detect e.g. strains of Zika, with potential point of care applications
Canada passed an anti-genetic discrimination law that went into effect on May 4, which means no-one can be required to have a genetic test or disclose the results of a genetic test, but PM Trudeau is challenging its constitutionality
Experimental evidence of long-lasting (14 generations) epigenetic memory of environmental change in C elegans

Round-up March 28th – April 3rd

I will be taking a month off from Round-ups — next install in early May.

In the next step of the CRISPR patent dance, the European Patent Office has announced its intent to grant the UC Berkeley crew a patent that would have broad coverage of CRISPR; though this will almost certainly be challenged.

Analysis of protein truncating variants from the ExAC group to identify which genes have the strongest selection against loss of function finds these have a strong overlap with genes known to be lethal in mouse knockouts.

A large scale joint analysis of Structural Variation and gene expression found that SVs are causal at 3.5–6.8% of expression quantitative trait loci (eQTLs) – a larger estimate than previously thought. Moreover, the effect size of SVs was larger than for SNVs or indels. Most of these SVs effected regulatory elements.

CiVIC is an open data project linking somatic mutations to particulae cancer types. Illumina has just announced that they will contribute over 8,000 associations to CiVIC, tripling the size of the database.

Variation linked to more production of the vitamin folate, and a higher BMI, have been found in Tibetan genomes, potential clues as to how they have evolved to thrive in an environment with 40% less oxygen than sea level.

Stanford researchers introduced Transcribed Genome Array (TGA), for probing transcriptome wide binding affinities using a hardware/software system based on the MiSeq.  “Our work couples transcriptome-wide measurements of binding affinity, sequence, and structural determinants of binding, and phenotypic outcomes to provide a comprehensive portrait of Vts [a developmental regulator] function.”

A win for the intersection of cognitive neuroscience and genetics, with the discovery that the expression of several genes is linked to brain activity during memory processing.

In some parts of the world, infection with the fungal aflatoxin causes 80% of liver cancers. A clear mutational signature of this infection has been reported, holding out promise for early detection and assessing exposure.

A GWAS of over 12,000 cases identified new risk loci of glioma suggest a polygenic susceptibility, and different signatures for glioblastomas and non-glioblastoma tumors.

5% of reading ability is attributable to genetics (compare 1% attributable to gender)

Repositive, a new service for aggregating genomic datasets.



Round-up March 21st – 27th

Last week was the American College of Medical Genetics’ annual conference. I’ve yet to find a good write up or summary anywhere!?

Where do cancer causing mutations come from? Traditionally, we have focused on inherited mutations, and environmentally caused mutations. But there is a third source, arising from errors in DNA replication. A study in science reports that this third source, which are unavoidable, are responsible for two thirds of mutations in cancer. This strong role for “bad luck” (rather than behavior) has previously caused controversy.

study in Nature uses the mutational signatures of cells of adults to elucidate facts about the very earliest stages of embryonic development. They report that each cell division results in three new mutations, and that the two daughter cells tend to contribute to cell numbers in a 2:1 ratio.

GSK and Regeneron have teamed up with the UK BioBank, to sequence the genomes of the 500,000 individuals who are part of the BioBank cohort, with 50,000 expected in 2017. The ten year old UK BioBank is a cohort of individuals with extensive health records, described as “the world’s most comprehensive health resource”.

The chromosomal level abnormalities underlying different types of acute myeloid leukemia (AML) have been known for a while; a large study correlates this with underlying mutational changes, allowing for subtyping the disease.

CAP and AMP publish guidelines for the validation and monitoring of NGS based somatic panels.

personal story about why it is so important that we do not allow GINA’s protections to be eroded, form a woman who speaks of “mutants like me”.

Making the case for the role of bioethicists in a healthy research environment.

As rounded-up last week, the results of the PSCK9 trial were generally perceived as underwhelming. But, Robert Plenge writes, as a proof of concept for genomic based medicine, the results were encouraging.

Variants that affect splicing have a good chance of being disease causing. Its often not possible to assess whether a variant affects splicing computationally; an experimental technique using a hybrid minigene based method.

study showing that recurrent pregnancy loss is associated with lower number of copy number variants in the placenta.

An information-theoretic approach to the epigenome.

variant in MAOA (one of the most controversial genes in the genome, sometimes called the warrior gene), already linked to alcoholism and smoking behavior, has been linked to heroin addiction.



Round-up March 14th-20th

The news this week has been dominated by Trump’s plans to cut the NIH budget by 20%. We’re also bracing for the coming storm of announcements/reported results expected from the American College of Medical Genetics conference, which starts in Phoenix, Arizona, tomorrow.

In the nature-nurture debate, twin studies have played a decisive role in helping tease apart the relative contributions of genetics and environment. A meta-analysis of twin studies covers ~18,000 traits and ~15 million pairs of twins, shows an average heritability of 49% across all traits.

Oxford Nanopore have announced the launch of a new desktop product, the GridION X5. It is five of its MinIONs plus a lot of compute in a box, and much smaller than the PromethION. Written up my Omics! Omics!.


  • A “good news” variant in PCSK9 is associated with lower levels of LDL cholesterol and lower chance of heart disease. A large scale clinical trial of a drug that targets PCSK9reported today that it did work, but not as much as analysts had been hoping for, and perhaps not enough to justify the $14,000 price tag.
  • More tumors than previously thoughtmay be BRCA1/BRCA2 deficient, meaning that more patients could potentially benefit from PARP inhibitors.
  • eGenesis, a spinout of George Church’s lab, has raised $38m in Series A financing. They aim to make pig organs transplantable into humans, using genetic modification to combat organ rejection.
  • A clinic in the UK has been the first to be given the go aheadto make three person babies. The UK recently made the procedure legal.
  • HudsonAlpha is offering an “elective genome”at $7000. 7 of the first 24 patients had actionable information reported. The focus is on rare disease, one only those genes associated to conditions that the patient has a personal or family history of. The test is in large part being offered because patients ask for it — but this isn’t necessarily good reason to offer a test.
  • A number of companies are trying to use DNA as a tracer molecule– an alternative to dyes or radioactive materials.
  • Making the case for building the infrastructure to report on protective variants.

New methods

  • Apaper showing that an antibiotic compound allows some cells to “read through” premature stop codons, giving hope to those who suffer from rare disease caused by such mutations.
  • A group has reported single cell level structural maps of the mammalian genome, at a resolution of less than 100kb. Some things are constant between cells (A and B compartments, lamina-associated domains and active enhancers and promoters), while some vary (individual topological-associated domains and loops).
  • Some tumors can be attacked by a combo of drugs that act synergistically. A CRISPR-based double knockout (CDKO) system, designed for high-throughput detection of which pairs of genes give a phenotype when knocked out, allowing synergistic drug target combinations to be identified. Another study of 142k gene-interaction testsreplicated combinatorial drugs at 75% precision.

New genetic associations



Round-up March 7th – 13th

Canada has passed a Genetic Nondiscrimination bill. Insurance groups opposed it. So did PM Justin Trudeau, saying it impinges on Provinces’ rights to regulate the insurance industry.

Meanwhile, a US House Committee approved a bill that will enable companies to inflict financial penalties on employees who do not submit to genetic tests, thus undermining GINA’s (Genetic Information Nondiscrimination Act) ability to protect the genetic data of employees. Employees who refuse testing for “workplace wellness” programs will face reduced pay, fines, or inflated insurance premiums. Tom Price, the HHS Secretary, has said there would be “significant concerns” about the bill, and that they’d be taking a look at it when it got to the department.

The Personalized Medicine Coalition has published a report on current opportunities/challenges.

On the opportunity:

  • Percentage of the patient population for which a particular drug in a class is ineffective, on average: ANTI-DEPRESSANTS 38%; SSRIs ASTHMA DRUGS 40%; DIABETES DRUGS 43%; ARTHRITIS DRUGS 50%; ALZHEIMER’S DRUGS 70%; CANCER DRUGS 75%
  • There are 132 personalized medicines on the market; ~42% of drugs in the development pipeline include biomarker investigation as part of R&D; personalized medicines accounted for 27 percent of new drug approvals in 2016

When it comes to challenges:

  • Lack of regulatory in landscape puts some investors off;
  • Reimbursement, where there is a Catch 22 situation “Widespread insurance coverage of diagnostic tests, for example, will likely require practice-based evidence demonstrating value. Obtaining the real-world data necessary for generating this evidence, however, is difficult unless the products and services in question are covered by insurance policies.”
  • Decreased payment rates, partly because the CMS moved from allowing “stacked codes” to a “gapfill” methodology, which allows regional contractors to set prices
  • A new rule that appeared in the Protecting Access to Medicare Act (PAMA) may put downward pressure on utilization of personalized medicine, as there is no mechanism for capturing the value of targeted treatment
  • Clinical Adoption is slow due to lack of education and lack of IT infrastructure — “most health care organizations do not have formalized plans to leverage advances in genomics and data analytics to personalize patient care, and are unprepared to implement personalized medicine programs”

Scott Gottlieb has been chosen by Trump to head the FDA. The most moderate of the names doing the gossip rounds, he is thought to support maintaining ensuring drugs are both safe and effective before approving them — some other candidates supported dropping the “effective” part.

Science news

  • An article in Naturereports on a new technique for uncovering the 3D architecture of the genome, including mulit-enhancer contacts.
  • Where do circulating DNA fragments come from? A new methodlooks at methylation haplotypes to determine tissue origin — allowing for cancer status and tissue origin to be determined from a blood draw. This could be a big deal in the liquid biopsy space. Singlera Genomics will attempt to commercialize the technology. Sensitivity can be increased by running another methylation test in parallel, based on the open/closed chromatin status.
  • The gene CDH2 has beenimplicated in sudden cardiac death.
  • The gene SEMA4D has beenlinked to obesity in African populations – those with variants in this gene were on average 6lbs heavier. Obesity is more common in those with African ancestry than others, and other populations lack variation in the gene. The study highlights how important it is to study disease associations outside of Caucasian populations.
  • A review on using genetics for transplants.
  • Alarge effort to sequence plants used in Chinese Medicine.
  • And on the subject of prediction, a model that incorporates omic-datato predict years of life post treatment for breast cancer is able to explain some of the variance after more conventional variables are factored in.

In the “healthy exome” space, Arivale have started offering polygenic profiles — i.e. giving their customers risk scores that aggregate over many variants. There is little data that speaks to the scientific validity of this.

George Church is teaming up with brain training game Lumosity to uncover the genetics behind those with outstanding memory.

Geisinger are leading the way with implementing genomic medicine. Here is a Mendelspod interview with their Director of Clinical Genomics.

There are dozens of direct to consumer tests that allow for determining paternity. This study looked at the privacy policies and terms of services of 43 such companies, and concludes that “recreational genetics carries both the risk of unintentionally revealing misidentified paternity, and also the risk that fathers will deliberately use these services to test their children’s paternity without revealing their intentions to the mother or any other third party.”