Hello, and thank you for listening to the MicroBibCade podcast. Here, we will be
discussing topics in microbial bioinformatics. We hope that we can give you some
insights, tips, and tricks along the way. There is so much information we all
know from working in the field, but nobody really writes it down. There's no
manual, and it's assumed you'll pick it up. We hope to fill in a few of these
gaps. My co-hosts are Dr. Nabil Ali Khan and Professor Andrew Page. Nabil is the
Head of Informatics at the Quadram Institute in Norwich, UK, and Andrew is the
Director of Technical Innovation for Liagen in Cambridge, UK. I am Dr. Lee Katz,
and I am a Senior Bioinformatician at Centers for Disease Control and Prevention
in Atlanta, United States. Hello, and welcome to the MicroBibCade podcast. I'm
your host, Andrew Page. I'm here with Lee Katz. And we've got a special guest,
Finlay McGuire. We are at the Global Microbial Identifier Conference in
Vancouver, Canada, and we have been set the challenge of being roving reporters.
So, hello, Finlay. Do you want to tell us about yourself? Thanks for introducing
me. I'm Finlay McGuire. I'm an Assistant Professor at Dalhousie University in
Computer Science and Epidemiology, which is right on the opposite coast of
Canada, a seven-hour flight away. But you're not Canadian. I'm not Canadian. I'm
originally from Scotland. I have permanent residency, but not quite citizenship
yet. As many people found for this conference, the Canadian Immigration Service
is very slow and just rejects people on occasion. Yeah, not very good for a
global conference. So, what I want to know from you, Finlay, is what's your
background? Are you a computer scientist or microbiologist or molecular
biologist? Where did you come from? I am a complete fraud in both of my academic
affiliation departments, where I have no qualifications in computer science or
epidemiology. I come from a microbiology background. I did undergraduate
microbiology and then a PhD that was meant to be 50% wet lab, dry lab work,
before I rapidly realized wet lab is the worst, and I wanted to get as far away
from it as possible, but entirely computational. And so, what is your actual
current affiliations? I'm an Assistant Professor at Dalhousie University in the
Department of Community Health and Epidemiology in the Faculty of Computer
Science. That's a mouthful. Yep. I'm also a Pathogenomics Bioinformatics Lead
for the Sheriff Hospital Lab, which is a large med micro lab based at Sunnybrook
Health Sciences Centre in Toronto. Okay. So, how on earth do you do that many
jobs simultaneously? Are you just one of these wonder people who works 80 hours
a week, or is it just how you have to do it to get funding in this country? I
was in a position where I was lucky enough to be finishing postdocs and got two
job offers. And I was like, I know, I'll negotiate with both parties and come up
with a job that's a fusion of the best parts of both. And that's how I ended up
with two separate jobs. And you're paid twice, yeah? Mostly paid once, but there
is some funding from the clinical lab, yeah. Now, Lee is very quiet here because
he's currently eating a cookie, but we'll let him off. Well, let's get into the
meat of it. We're at the GMI Global Microbial Identifier Conference. How did you
find yourself here? Thanks to Dr. Emma Griffiths, who is my co-chair in the
Phage Data Structures Working Group and one of the organizers of the conference,
basically invited me along. I'd kind of missed GMI, I think, by being more in
the purely public health side rather than the broader global data sharing of
microbial data. And so what are your actual interests academically day to day? I
know you do a bit on data standards and you do a little bit on AMR. So where's
your real passion? Basically making genomic data usable for clinical decision-
making, real-world actions, as well as some of that public health, kind of
driving public health policy and decision-making and sort of developing tools to
try and support that. But really the whole gamut from individual patients to
global health kind of policy. That's a very big gamut. Now, do you do animals or
just human? I do animals as well. So we recently identified, or recently, last
year, we found SARS-CoV-2, which had jumped into deer, then back into humans in
Canada. That's nice. So we let that work, so a zoonotic virus. Was there many
SNPs when it went back in? 76 new mutations, highly divergent. Jesus Christ. It
was like a massive long branch. We actually identified the human case before we
found out about the deer, because there was a human case that was massively
divergent from anything else. Yeah. That's really good work, actually, you know,
and all these hidden links that we don't necessarily normally see that go on
within nature and the environment. Yeah. So you're obviously integrated in this
community with GMI, with Phage, and we just had the first few sessions here. I
guess I won't name them all here, but what struck you the most? What are you
thinking about? What do you think is unresolved right now? I mean, as again, as
kind of mentioned a few of the questions, like the devil's always in the
details. Like I think, you know, the guiding principles for data sharing with
WHO actually did integrate a lot of kind of comments and recommendations really
well and come up with, yeah, those guiding principles, but that kind of
operationalization step that, like, we've seen with, you know, Malban's work
with Josephina and some of the PulseNet kind of work, but kind of extending that
more broadly, especially to that One Health conception, which has become a
critiqued word, if not a meaningless buzzword in some contexts. It has been
abused. But really it should be even more broad. Like we have things like we
have plant pathogens as well, which have huge health impacts, which have a lot
of the kind of same mechanisms of resistance that we see in antibiotic
resistance, antimicrobial resistance, that is a completely separate ecosystem.
It should be part of that One Health discussion. Plants are hard though, you
know? Why would you want to do that? Yeah, but the pathogens are plants. True,
true. Bacteria, viruses, and a lot of tricky fungi. Yeah, yeah, yeah. And like
nematodes. I do suppose that we're pretty Foodborne focused this morning.
Actually we are, yeah, but that's okay because I come from Foodborne background,
so yeah. Yeah, okay with me too. Okay, but that's good noticing that, so. I feel
like Foodborne was like really one of the drivers of WGS. Big time, yeah.
Particularly in the U.S. Sure, I mean even a couple of the speakers said that
they follow the PulseNet protocols even though they're not part of the U.S. And
then obviously Genome Tracker and all that kind of project getting sequencers
out into all the state labs in the U.S., that was pretty good, you know? And set
them probably 10 years ahead of everyone else. Yeah. I guess I won't comment too
much on that since we're at GMI and I want to focus more on the guests, but
yeah. It was interesting actually the different models for data sharing that the
WHO had, you know? And it was very clear one was very aimed at GISAID, like it
was like the GISAID model, but they didn't say it. And then the other was the
INSTC model, which is, you know, share everything with no restrictions. So, you
know, what's your opinions on those two different sharing models? And, you know,
obviously, you know, you're not Steve Myers or anything like that. No, I did
receive a phone call from him. Wait, are you Steve Myers? No, but I did receive
a phone call from Steve Myers early in the pandemic. A lot of the concerns that
GISAID was created to alleviate are important concerns. Where there's a bit of a
challenge in culture is it's a bit like that kind of public health and sort of
equity sharing data ownership model is very much at odds of what had been the
prevailing momentum within genomics of entirely open data, open tools, open
resources without any restrictions. And so kind of trying to square that circle
and resolve those two different sets of concerns is a challenge and a thing that
we're actively discussing a lot. What I find really interesting in terms of data
sharing and access is crossing that bridge between human and agricultural is
often the agricultural data is way more locked down and harder to share than the
human clinical data is. Which is crazy. You wouldn't think that, but actually it
is, yeah. Well, and it's basically because there's a lot more, you know, there's
a lot more laws for the human side, but there's a lot more lawyers in the
agriculture, agri-food business. Yeah, and you can see when it goes wrong, like,
then it's a big problem. Like, what was it? Baby formula in New Zealand years
ago where they taught they had, was it botulism anthrax? One of those in the, I
think it was bot actually, in baby formula. And then it's a problem. And so they
dumped, you know, like a billion dollars worth of baby formula. And it turns out
actually, you know, it was just a lab test, couldn't get them to resolution that
they needed to detect bots. Yeah. That's incredible. I know. And really they had
no positive controls within the country and things like that, you know. And
then, this is, I'm going to hear this thing about talk. The people, they did
sequencing, but then they didn't have the required machines to analyze the data.
And then they had to kind of jerry-rig stuff and they weren't allowed to move
the data around and all these kind of complications, which just led to delays.
Yeah. Hopefully all fixed now. Yeah, fingers crossed. One thing that I find
interesting on like the food contamination angle is, obviously there's a lot
more forms of food contamination than microbial. So like, where's the
intersection of data exchange with like other forms of food contamination, like
within organizations like the FDA? Is it a completely separate set of labs that
are doing like mass spec and other kind of assays for that? Pretty much, yeah.
There are things like that, I know that FDA has their inspectors, and then they
have CIFSAN, and CIFSAN will do the whole genomics workflow, and they can do
that, so I know that's divided. And then we have FSIS for agriculture, so
everybody has their own specialty and their domain. It's similar in the UK, it's
quite split up, and then there's different government committees for each
different type, you know, if it's chemical contamination or if it's microbial or
whatever. So yeah, I guess it'd be pretty bad if you had multiple different
types of contamination in one food source. Yeah. Very, very bad. Well, years
ago, there was the tuna scrape outbreak in 2012, I believe. Is that the sushi
one? Yeah. And from what I heard, like basically third-hand, not so this isn't
primary, but like what I heard is like people in India were like using raw clam
shells to like scoop out tuna, and there are just like so many different ways
where that could go wrong. Yeah. And it's a wonder that we only had to get FDA
involved, but it's like how many different agencies would have gotten involved
if more pathogens were there? Yeah. I mean, this is going back to your well-
traveled salad, you know, do you remember those? Yeah. The well-traveled salad,
do you not remember that? For about a year, every single presentation in our
area had this picture of a salad, the well-traveled salad showing all the
different bits of the salad and where it came from in different countries. You
know, you could have had sweet corn from like three different countries in your
one little salad. You could have had, you know, tomatoes from one place, lettuce
from another, peppers from another. You know, where does contamination come in
from? When you do have a problem, well, you know, it's very difficult to track
it down, even with sequencing. We had the example also in the US, it was like in
a nice New York Times article in 2009, the hamburger that comes from all the
different places in the US or in the world. I mean, the UK had the hamburgers
that didn't come from cows. Well, at least horses are safe. It was safe for
human consumption. It's, you know, the mad hamburger is, that's a problem. Oh,
yes. You know what I say about horse is, nay, nay, sir. Well, it's very
expensive in France. Like you have specialist horse butchers, you know, and it
is quite a premium. But I think the horse meat that has been used in this case
was very low quality, you know, old nags. And actually, I recall it was
discovered by the Irish food safety people at the end of the year. They had a
bit of money left over and they wanted to spend their budget. So this guy, just
go out and buy some ready meals, whatever, and in a few supermarkets and they
did. And then they went and sequenced them and then they found, oh, it's not
beef in this cheap lasagna from the supermarket. It's actually something else,
et cetera, et cetera. And it was just for Christmas and then it all broke and
everyone was very upset, particularly since a foreign country had detected the
stuff in the UK and not the UK, it's safety authorities. But there you go.
That's how it is. Let's bring it back to GMI. Oh, sorry, GMI, yeah. Yeah, yeah.
So GMI is like phage except different. Yeah. There are some, I think there's
still kind of an active process kind of going on to kind of delineate and do a
bit more. There was a lot of discussion about having a landscape kind of
evaluation about all the different groups and organizations kind of active in
this space and the dividing and conquering of what people are actually doing and
covering. I do think phage, depending on the working group, does a good job of
doing a lot of boring but useful specific stuff, trying not to get too lost in
the weeds of broader global strategy and focus on, okay, how do we have a
pipeline with more than one AMR detection tool in it? Yeah. And actually putting
it into something useful like a Python, you know? Yeah. So you can actually go
and use it, not just have a document that people will read and file away after
they weigh it. What metadata should you collect and put it, and how do you put
it in a database? Actually describe it in a spreadsheet so that people actually
understand and then have to look up tables between different things that people
use. That's actually really useful stuff, and I think that's the practical side
that can be missing from some of these international collaborations. But Devil's
Advocate, why shouldn't GMI be doing that? I mean, I think they have done in a
lot of spaces, like in the past, but I think they also were doing a lot more of
that kind of strategic, global, almost more politics angle as well, certainly
some of the working groups, which is a whole set of challenges. I'm really doing
Devil's Advocate because I've been in that technical working group for GMI also.
Which working group's in? The standards and data analytics part. And are you in
the FH rival one, are you? I'm technically in there, but I'm inactive. I would
say I'm definitely inactive and skipping all the meetings and telling Emma to
stop including me in authorship. She's very inclusive. She is. She's a very good
coordinator. So the big question is, right, Tim Hortons, what is your favorite
donut? Our donut, Tim Hortons. Because we are in Canada, you know? Wait, wait,
wait, wait, wait. Do you like Tim Hortons? That's okay. Okay, all right.
Honestly, if I go to Tim Hortons, I often get the bagel and cream cheese more
than the donut. Oh, okay. What would you recommend to us foreigners who are
obscene being a local? I mean, you know, Timbits are the usual classic go-to in
the sweet end, which are just little donut balls of various types. You can get a
mixed set with different sprinkles and flavors on them. That way you can sample
lots of different donuts. They're often a feature of our epi department coffee.
So 50 Timbits, there you go. I walked there yesterday and got a Boston cream
donut. What's your reaction? Knock yourself out. Not very Canadian, actually, if
it's Boston. I know. I do remember an undergraduate, apparently, like Tim
Hortons, like opened, you could buy it in a spa in the UK for a brief period of
time. Oh, okay. I remember like one of the Canadians get very excited and us
having to get the train out and like did call or Banbury or something to go to
the spa in the gas station to buy Tim Hortons donuts. That's yeah. It was not
worth the trip. Yeah. Sounds kind of gross. I mean, Sbarro. We digress. Well, we
ended there and we will come back with more updates later from the conference.
Yeah. All right. Thank you very much for that little discussion. And so we're
here talking to Emma Griffiths, who is one of the organizers of this lovely GMI
conference and has just fed us very, very well. Thank you very much, Emma. And
actually, we're in a very beautiful room here at the moment, it looks like,
where, you know, if you're an evil genius, you have all of your, you know,
minions around in a circle. That is the plan. That's the plan for this meeting.
Absolutely. World domination. Anyway, so why can't ChatGPT replace ontologies?
God damn it. This is a question that I asked last week at a conference. And then
left. Because a lot, you dropped the bomb and then left. I refuse to answer
these questions right now. Too stressed? Yes. Also, I don't even know what to
say. I have a question for you. Okay. I went to Tim Porter's last night. This is
the kind of question I can answer. Good. Right. Tell me if getting a Boston
cream donut was the right decision last night. Oh, wow. This, I don't, maybe I
want to answer the ChatGPT question more than the Boston cream. Now, I think we
talked about this on Slack, and a wise man once said, you know, your preference
of donut really depends on your mood and your circumstances. So, really, Lee, I
would have to ask you, what was your mood and your circumstances? Did the Boston
cream feel right for you? Because that would not be what I would have
recommended. Can I tell you what I did before? I think you should. Before I got
that, thanks for the microphone. Okay. Before I got that Tim Horton's donut, I
stopped by Jappadog. Fantastic. Did you love it? I loved it. I loved it. I got a
Wagyu beef with, like, some mayonnaise, onion, seaweed topping. But, like, from
a cart, right? Not from the shop. I got it from the shop. Oh, no. You did it
wrong. Damn. You have to get it from the cart. Okay. But I'm glad that you
enjoyed it. So, that's good. That's great. Well, I walked outside as though I
got it from a cart and ate it on the sidewalk. You really need to get it from
the cart. That just means you're going to have to do it again. And then I got
the Boston cream donut after that. Right. They have poutine in the restaurant,
too. Yeah. Oh, I don't know if I can vouch for Jappadog poutine. But poutine is
poutine. Well, that's a controversial thing to say, actually. Okay. But it's
better to have some poutine than no poutine. I sort of remember, this is coming
full circle.  I sort of remember Jappadog being somewhat of a weird flavor
profile and maybe it was created with AI. And so... Fusion. Oh, it's Fusion.
There were weird flavor profiles back then. I think Jappadog might have got
inspired by it. I'm trying to remember the news article. But anyway, why
shouldn't we do GMI with ChatGPT? Gentlemen, I am dropping this mic. Taking no
more questions at this time. Thank you very much, Emma, and we'll talk to you
soon. Thank you so much for listening to us at home. If you like this podcast,
please subscribe and rate us on iTunes, Spotify, SoundCloud, or the platform of
your choice. Follow us on Twitter at microbinfee. And if you don't like this
podcast, please don't do anything. This podcast was recorded by the Microbial
Bioinformatics Group. The opinions expressed here are our own and do not
necessarily reflect the views of CDC or the Quadram Institute.