Sunday, October 01, 2017

Dispatches from CDC AMD Day 2017

I had the singular honor and pleasure of speaking this past Monday at the Center for Disease Control and Prevention's Advanced Molecular Detection(AMD) program's annual confab in Atlanta.  Just visiting the CDC campus was already a bit magical -- along with the Kennedy Space Center and Cold Spring Harbor it's one of mythical places of human exploration to me.  But to actually stand at the podium? Wow!

I've collected below a bunch of separate mental threads, many of which probably should be expanded out to a full post in the future.

It Pays To Read the Conference Agenda

My gig came about because I've gotten to know Duncan MacCannell from AMD via Twitter.  He had asked if I'd ever consider speaking at the CDC and I was entirely interested -- the CDC is a bit legendary for me so even just visiting the campus would be a thrill.  AMD Day fit my personal schedule this year so came to an agreement.  I was a little nervous talking about infectious disease diagnostics, given that I don't work on the topic, but Duncan suggested I could do almost a verbally delivered blog post.

I'd been wrestling with structuring the talk a bit this past summer, without much progress.  When the agenda came in from a CDC staffer, I first passed it over.  But then it occurred to me that I really should see where I was in the program and how long a time slot.  First, I wouldn't want to prepare 20 minutes of slides for 15 minutes of talk.  Second, I might want to tune my presentation to those around it, perhaps latching onto the theme of the session.  Third, I'm at best a facultative morning person, so if I was scheduled early on the second day it wouldn't hurt to start steeling myself for that.

So I looked and at first did not find.  Where am I?  These all look like poster sessions.  Finally, it occurred to me to scan systematically, which I did from end to start.  Gulp! There I am with forty minutes following the opening remarks of the entire conference.  

It was a lot of fun and I think it went over well -- people who didn't have any need to say nice things about it did so.  Being the keynote also had a strange relaxing effect. I've seen a lot of keynotes done by people peripheral to a conference focus, so I didn't feel like an imposter anymore -- or at least I would be an imposter in an imposter's role.  Plus, while I work a joke or two into most talks, with a keynote I wouldn't feel any obligation to be excessively serious.  So Miss Amanda appeared on my slides, I had a slide lining up a Maxam-Gilbert sequencing pattern with an Apollo Command Module orbiting the moon (both used hydrazine!) and remarked that Kate Rubins wore some unusual Personal Protective Gear when traveling to her sequencing lab. A deck that name-checked Mick Watson, Nick Loman, Josh Quick -- as well as William Henry Harrison and Zachary Taylor. A slide titled "Snakes Alive!".  An image of ENIAC standing in for bioinformatics computing. Plus poked fun at my second scientific publication, an exercise in Monday-morning annotation of the first bacterial genome sequences.


A Prop Is Worth A Thousand Words

Given that my talk was titled "Rapid Field-Deployable Sequencing: An Armchair Strategist's View" and had a significant focus on the Oxford Nanopore MinION, I would have been foolish not to bring my MinION along.  Alas, I didn't think to save a washed flowcell from the last batch of returns, so it was an incomplete prop.  But most people who asked me after the talk hadn't seen one, so even just whipping the device out of my pocket was useful.  I won't claim I don't enjoy the looks of wonder when someone first sees one.

Skydiving's Starting to Look Pretty Safe!

AMD Day is the most focused scientific meeting I've been to in a long time.  Other than a discussion panel I'll cover, every talk or poster was on detection of infectious disease.  Other than a handful of us -- myself, a few vendor reps and a startup person or two -- everyone was either from the CDC or from a state public health agency or was an academic working with a public health agency.

The downside of this focus is that it could lead you to swear off a wide range of life activities.  Foodborne illness. Drinking water. Venereal disease.  Brain-eating amoeba acquired by swimming in fresh water.  Mosquito-borne disease.  Air conditioning.

Now, I was aware of all of these prior and most can't be dealt with by prevention - good food safety, proper water treatment and sewage disposal, safe sex.  The brain-eaters are a general worry, but are luckily rare.  The one that really got me nervous is Legionella.  It's not because I remember the intense coverage of the original outbreak at the American Legion convention in 1976 -- growing up outside Philadelphia this was huge, front-page news.  No, what is terrifying is that, according to a CDC Legionella veteran I chatted with at her poster, nobody knows what triggers outbreaks.  Legionella is endemic in water systems and air conditioning systems and usually keeps quiet, but for unknown reason strikes out to cause outbreaks.

This group was also very concerned with the unfolding natural disasters in the Caribbean and Mexico: food supply disruptions, drinking water systems broken, rampant puddles for mosquito breeding, disarray of the medical system.

On the lighter side, it is clear that some of the researchers in this space stay cheery with morbid humor.  A slide on an norovirus poster had a silhouette of a person on all fours, ejecta spurting from both ends.  One of the foodborne illness groups was offering candy to encourage visiting their poster.  And then there was Legionella Larry, a presentation on this bug delivered as a series of tweets by a bacterium. 

Public Health is Spectacularly Multidisciplinary

While the meeting was more homogeneous than most, the public health crowd is composed of a fascinating mix of training backgrounds.  I met docs who were infectious disease specialists but also former pediatricians. Nurses and medical technologiests. Molecular biologists and microbiologists and bioinformaticists obviously, but also individuals from the social sciences.  As well as individuals trained in the social sciences.  

The Training Challenge

Conversely there was a panel group on challenges of converting public health labs over to sequencing-based methods.  Training programs have emerged so that established workers can be brought up to speed on sequencing, bioinformatics and phylogenetics.  A somewhat quietly expressed fear: once so trained, staff may leak off to biopharma or even Silicon Valley.


Sequencing Technology Landscape

Given that the meeting was on Advanced Molecular Detection, it was interesting to see the mix of technologies employed.  I wasn't clever enough to actually tally the posters by technology, but I did go back through the abstract book.  The term "NGS" or its expansion reigns supreme, generally meaning "Illumina" but not exclusively.  About five times as many abstracts mentioned an Illumina platform specifically as did Ion Torrent, and most Ion Torrent abstracts were comparisons of data quality with Illumina.  PacBio wasn't quite as prevalent but had a good showing, and the sentiment that it is the best route to closed genomes was common.  A handful of posters discussed MinIONs, and I think they were all from the same group.  But a number of groups were excited about trying it, and there was very strong interest in the direct RNA kit for some of the RNA viruses.  As one might expect in a group with "Advanced" in the title, what little showing Sanger sequencing had was as a legacy technology to be compared against.  I remarked to a couple of people the GnuBIO (now BioRad) sequencing technology could be interesting for some of these amplicon panel efforts -- should that technology actually ever hit the marketplace.

Most posters were focused on sequencing, but there were a handful developing new qPCR panels.  Between this and setting up amplicon panels for sequencing, Fluidigm and Wafergen both made appearances.

Conversations uncovered technology that wasn't covered in the posters.  I also found out this way that OpGen has discontinued supporting or producing kits for their Argus optical mapper, a news item I missed.  So for AMD, optical mapping is synonymous with BioNano Genomics.  Some pilots have been run on 10X Genomics, but the lack of support on that platform for metagenomics, which was seen as the prime application, has meant only a small effort.  Dovetail and Phase Genomics are seen as interesting areas to explore.  Digital PCR was another tech that wasn't on a poster, but based on a conversation it is in the mix somewhere. 

A Thousand Flowers or My Way/Highway?

A number of posters illustrated a dilemma in any field: experimentation vs. standardization.  Many groups are trying to accomplish the same or highly similar tasks, yet there was often a wide range of methods employed.  For example, many of these labs are sequencing isolated bacteria with the first goal of nailing down the miscreant bug's genus and species.  So there are pipelines set up to assemble the raw data and then perform taxonomic identification.  Many of these pipelines have very similar or perhaps essentially identical goals, and many similar pipelines have been developed by academics outside of the AMD circle.

So is this wasted effort or useful innovation?  That is always a challenge to assess.  I will fully confess to being someone who only sometimes resists the urge to invent my own version of existing software.  That gives me more control and power (and sometimes is needed to avoid licensing tangles), but it also means spending time developing and debugging.

With public health pipelines, the stakes are obviously higher.  Not only could a buggy pipeline have grim repercussions, but an inability to compare summary results between labs could lead to great redundancy as each lab reruns the others results.  Conversely, determining which pipeline is best, even in the absence of any personal or political considerations, can be a challenge.  Proper "bake-offs" are significant effort to develop.  When differences are identified between pipelines, which matter?

Still, quite often such comparisons do identify a small number of clearly superior contenders versus a number of weaker foes.  Options or flags can be contemplated to give control within a pipeline, perhaps trading off speed and sensitivity.

A model which hasn't been replicated nearly enough is Nucleotid.es, a site which performs regularly updated comparisons of de novo genome assembly.  By using containerization via Docker modules, the site makes it straightforward for developers to upload their pipeline to be added to the comparison.


The Longitude Challenge

An interesting corollary to the challenge of standardized methods is maintaining analytical continuity so that long-term studies can be maintained.  For example, with Bordetella there are specific Variable Number of Tandem Repeats (VNTRs) which have been long used for fingerprinting strains.  It is valuable to maintain continuity with the older studies, so typing VNTRs by sequencing becomes important.  Apparently in these organisms the length of the VNTR is affected less by repeat expansion/contraction and more by decay of the sequence in the terminal elements of the repeat array.

In a related vein, one paper was looking at the recent yellow fever outbreaks in Africa.  A bit of investigating of the data showed an issue with the strains in Genbank -- only a small portion of the sequence is available for what is the most likely match.  

Conference Format Innovations

AMD Day was quite different from most meetings I've been to, being extremely focused on posters.  Indeed, there weren't any standard format long science talks.  Instead, each poster session was introduced with at "Speed Dating" session -- each poster presenter (and they were, it was nice to see, mostly junior scientists) was given one slide and one minute to present.  In the upper right corner a little pie was filling in and at the end of the minute the slide auto-advanced to the next poster.



A handful of poster presenters were given a slightly longer short format.  In this, they had a fixed number of slides -- set to auto-advance at quick intervals.  I think it was every 30 seconds for five minutes.

A bit of whimsy at the conference was a tie-in to a CDC Public Service Announcement campaign publicizing the role of AMD in protecting public health.  A green screen photo booth enabled participants to be composited into one of the campaign's posters.  Here I am with AMD Director Dr,. Gregory Armstrong M.D. extolling the virtues of pathogen sequencing.

Scientific Storytelling

A supreme treat of the conference was a panel discussion on Scientific Storytelling featuring Betsy McKay,  Susan Swanberg and Glen Nowak.  McKay is the longtime Atlanta Bureau Chief for the Wall Street Journal, so her domain includes the CDC.   Swanberg was once a bench researcher, then a journalist and now teaches scientific journalism at the University of Arizona.  Nowak served as the public relations director at the CDC and now teaches at University of Georgia.

The discussion ranged over an array of topics.  For example, there is the distinction between news and stories.  Scientific news is focused on breaking information and will is much more likely to be numbers rich and "just the facts". Stories are meant to connect readers to a story and to use literary techniques -- they have narrative arcs and reveal the personalities of participants.  For example, in the context of the devastation of Puerto Rico, a news piece would focus on statistics (X% of the population is without clean water) where a story might profile a medical team struggling with the workload in the face of utility and supply disruptions.

I was the first at the question mikes and when there wasn't someone immediately behind me got to get a second question in.  Honestly, I would have paid serious money to take these three to lunch.  It's endlessly fascinating to understand how journalism really works and also to hear veterans of the field explain how they chosen to structure a given story.  For example, McKay described a story she wrote titled Rural America’s Childbirth Crisis: The Fight to Save Whitney Brown.  A key decision was how to structure the story in terms of the outcome -- lead with the patient's fate or save that to the end.  She also detailed how she kept circling back to the patient's family for more information, as when she pieced together accounts from other participants there would be inconsistencies or gaps.  Each time a bit more information came out as the family became more comfortable with the reporter and with revealing uncomplimentary information about the patient.  Swanberg cited Mrs. Kelly's Monster, which won a Pulitzer, as another story well worth studying.

Having panelists from each side of the news business gave this panel additional richness.  We could hear from McKay or Swanberg how she might treat a story about the CDC versus how Nowak would be thinking about the same topic.

Gonna throw "Course in Scientific Journalism" on my bucket list!

CDC Museum: A Cloaked Gem

Part of the conference building hosts the David J. Sencer CDC Museum.  It's unusual to need government-issued IDs and a walk through a metal detector to get into a public museum, but it is open to all who can meet those requirements.  Run in conjunction with the Smithsonian Institution, the permanent exhibits tell brief histories of the CDC and public health, recounting both some of the high points in the search for HIV and Legionella as well as not hiding the ignomy of the Tuskegee Syphilis Study.  Artifacts include a transmission electron microscope, pamphlets and posters from campaigns.

The upper level and mezzanine currently host an exhibit on the 2014 Ebola outbreak in West Africa.  It's very information-dense and covers all the aspects of the crisis, primarily through static displays but also through video interviews and playing the Frontline documentary.  An ABI DNA extraction robot is there along with examples of personal protective gear.  Again, the exhibit doesn't shy away from the ugliest and most difficult aspects of the crisis -- the slow response by the developed world, the distrust and mistrust engendered by prior colonial rule, the public and political hysteria over patients arriving in the U.S.  But it also covers how solutions were found to the scientific and cultural issues.   An excellent exhibit; I sincerely hope they develop a traveling version as this story should be told far-and-wide, and to be honest it was head-and-shoulders above most biomedical exhibits I've seen.  Alas, one thing I liked was the emphasis on detailed labels and artifacts whereas museum exhibits have swung far towards dubious interactive experiences, so I doubt a traveling exhibit would be quite as good as what is in place in Atlanta.

Well, that's all for now -- except one last picture from my fruitful visit to the CDC.  Plus many thanks to the CDC staffers who made this experience possible through their work in planning my trip and running the conference and taking nice shots of me at the podium.


1 comment:

dr7ski said...

Nice post Keith. About your comment "Training programs have emerged so that established workers can be brought up to speed on sequencing, bioinformatics and phylogenetics. A somewhat quietly expressed fear: once so trained, staff may leak off to biopharma or even Silicon Valley", I have a different take on this. I would rather train the staff and have them leave, than not train the staff and have them stay. I try to keep them intellectually and emotionally vested in the public health research they are performing hoping that they choose to stay. As one of my first mentor's once said, you need to find research that "galvanizes your soul".