20 THINGS I LEARNED DURING #FOGM17

doc.ai
doc.ai
Mar 7, 2017 · 4 min read

10th Annual — Future of Genomic Medicine —

March 2,3, 2017 | La Jolla, California

1. Cognitive decline is synapse loss (‘use it or lose it’). Microglia are the sort of bio-bots that eat synapses with low page rankings.

2. The only generative ML some researchers seemed to use was Bayes. No NN at all because ‘doctors want to understand the causative factors and NNs are black boxes’. Please correct me if I am wrong.

3. Clinical genomics is going for 7 days turnover instead of 3 weeks; Scripps is going for 26 hours. It’s a gene race.

4. In genomic medicine the only thing that really went mainstream in 2016 are ‘some of the words’ (dixit Eric Topol). The same in Machine Learning and Natural Language. We are all reading papers now.

5. Should we CRISPR human embryos or go earlier and screen and edit the hell out of germ cells? We could optimize humanity with a gradient descent :)

6. The Broad Institute alone has 254,000 fully sequenced genomes and some 68,000 exomes, which must constitute 33% of global sequencing. In 2016 they sequenced more genomes than in the previous 7 years (since 2009).

7. It remains surprisingly difficult to transfer data from one place to another. (No kidding!)

8. Some people do not want prediction. They prefer deliberate ignorance: Cassandra’s regret: The psychology of not wanting to know.

9. Silicon Valley startups were defined as: iterative, learning, disruptive, ambitious and well-funded. Yep, we are not just here for the good weather.

10. HAIL!!!! Now we are talking! Hail is an open-source, scalable framework for analyzing genetic data. Just VCF (https://github.com/samtools/hts-specs) is ok, everything is in Python, built on Apache Spark to efficiently analyze gigabyte-scale data on a laptop. We can even get involved in development: Github repo, and you can chat with the crew in the Gitter dev room. Carbon-based units, not bots.

11. A lot of the audience had gone through a full sequence genome (or exome) or SNIPs and they were divided into two groups: those who had found it interesting and those who had not. For those it was not interesting, it seemed to be linked to the fact that they were boringly healthy. Those who had mutations, were hoping to beat the odds with 3 out of the 4 musketeers of salvation: diet, non-smoking, activity, BMI.

12. Circulating tumor DNA is not always pathologic. As long as cancer is about the journey and not the destination, it seems to be ok.

13. The participant will see you now (seller’s market of research becomes a buyer’s market)

14. NIPT is the fastest growing test in the history of medicine (1 out of 4 pregnancies predicted in 2017). No denialists here.

15. We seemed to have pivoted our ambition with a new nomenclature in 2016. Long-term remissions are now called ‘cures’. Healthy aging means the conservation of the cognitive function.

16. In 2016 the death rate frequencies from cancer did not move the needle: 2 out of 10,000 improvement. A chance to live longer means 3 months and costs $100,000

17. We are using statins as candy, we might as well put it in the water supply (a wet dream for Pfizer et al). This was Eric Topol, only he can say that.

18. Japan wants to be the leader in genomic medicine and promises accelerated approval J

19. IPSC (Induced pluripotent stem cells) can now be used in reproductive scenario’s: from skin cells to germ cells, a new source of gametes for IVF. They can even be the source of germline gene editing: if it hurts pluck it out of the human genome.

20. The GC-inevitability rule: you cannot go to a genomics conference without somebody on stage mentioning the name ‘George Church’ (I have a data set that proves it). This time his recipe for superheroes (edit in protective gene variants)


doc.ai was one of the small sponsors of the Future of Genomic Medicine this year


About the Author Walter De Brouwer , Cofounder-CEO doc.ai

walter@doc.ai

My interests are in Machine Learning and Natural Language Processing. My background in genomics is recreational. My PhD (The Biology of Language) studied the relationship between Turing Machines, Transformational Grammar and DNA.

doc.ai

Written by

doc.ai

The AI powered medical research companion.

Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch
Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore
Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade