Hello, and thank you for listening to the MicroBinFeed 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 writes it down. There is 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 Dr. Andrew Page. I am Dr. Lee Katz. Both
Andrew and Nabil work in the Quadram Institute in Norwich, UK, where they work
on microbes in food and the impact on human health. I work at Centers for
Disease Control and Prevention, and am an adjunct member at the University of
Georgia in the U.S. So hello, and welcome to the Microbial Bioinformatics
podcast. Nabil, Lee, and myself are your hosts for today, and we are joined
again by Professor Mark Palin, who is a Professor of Microbial Genomics in the
University of East Anglia and a Research Leader at the Quadram Institute. We're
continuing our trip down memory lane and reflecting on some of the exciting
bioinformatics moments and lessons learned for bioinformatics in the 21st
century. So let's continue our conversation with Mark. Now then something
strange happened. I'd kept in touch with Nick Loman over several years. In fact,
he'd even helped me set up my server when I arrived in Belfast. And I'd applied
to get funding to build more databases on the back of our efforts with CoDubase
and KNPDB. I'd had a smart guy called Roy Chowdhury been working on this
CoDubase for many years. And I got this funding for a project that we called
X-Base, which was to apply that sort of thing across the board to all sorts of
genomes. And so I was looking for a smart bioinformatician, and I said to Nick,
I said, look, I've got five years funding here. Why don't you break out of your
medical career and come and do a PhD? You've got plenty of time in those five
years, you come on the payroll of staff, so you wouldn't have to pay PhD fees,
only like 500 for the whole thing. And you can return to medicine with a PhD.
And it turned out that he said, well, now is the right time to jump out of
medicine because the UK government made a complete mess of career progression
for junior medics at that time. There were far more people in training than they
provided jobs for. And so Nick said, yeah, all right. So I persuaded him to come
and work for me again. And so he moved over to Birmingham. And just around the
same time, that next generation sequencing, or high throughput sequencing as we
might call it, perhaps more accurately, came into view. And so that was a really
exciting time as we suddenly, we started looking at Moore's law applied to
sequencing and thinking what in 2025, which is when I'm 65, how much did it cost
to sequence a genome? I remember we got a ruler, we said, oh, it costs 25 pence.
And we were just joking, it can't be true that you could sequence a bacterial
genome for 25 pence. But I think we will be able to sequence a bacterial genome
for 25 pence in a few years' time. It's down to a few pounds. But we were there
at the beginning and we got involved through a collaboration with George
Weinstock. I'm still talking to all my clinical microbiology colleagues, I still
had a clinical academic role there. And we had particular problems in an
organism called Acinetobacter baumannii in the hospital. And I said, maybe we
could track the spread of this thing by sequencing its genome and see how they
relate to each other, sort of draw an evolutionary tree of the strains within
the species that come out of our hospital. It might work, it might not, let's
just have a go. And so we did that, George sequenced three isolates, we were
able to call out some SNPs between them. And it was the first paper looking at
gram-negative genomic epidemiology. A very smart person called Sharon Peacock,
who'd actually been on the same scheme that Gordon Dugan had done for PhDs for
medics, had actually got into the Sanger and pipped us to the post doing genomic
epidemiology of Staph aureus, they had a paper out. But we were the first ones
to do it on a gram-negative, and we ended it with three, but it was proof of
principle. And we then went on and did a much larger study a few years later,
where we tracked the spread of an outbreak strain and sequenced over 70 genomes
from patients and so forth. But that was a thing where we started getting into
outbreak genomics and looking at what we could deduce from genome sequencing by
using that approach. And it just so happened around 2010 or so, this new wave of
progress came in where it wasn't just next generation sequencing, it was
benchtop sequencing became a possibility. And three platforms came to market
around the same time. So there was the 454, 454 junior, that selects the
sequencing and transmogrified into Illumina sequencing and they had the MySeq.
And then there was this newcomer called IronTorrent that came on. And
IronTorrent launched a European-wide competition saying that if you win this
competition, you can have an IronTorrent for free. And it belongs to you as an
individual, not your university, you as an individual. And I said that we could
use IronTorrent to sequence our acinetobacters and track the spread of this
deadly multidrug-resistant pathogen. And they swallowed my Blarney and gave me
an IronTorrent. I won an IronTorrent, I went to La Salle to collect the award
and I got an IronTorrent. I said to Nick, you know, nobody's ever used this
thing before, see if you can work out how to handle the data coming off of it.
And so we started playing around and said, yeah, I think I've got the pipelines
working. And just at that time, there's this terrible outbreak in 2011 in
Germany of E. coli H4, sugar toxin-inducing E. coli in Germany. A group in
Germany collaborated with the Beijing Genomics Institute, BGI as it was called.
And they had got an IronTorrent and they sequenced the outbreak strain on an
IronTorrent and released the data into the public domain. And so I said to Nick,
I said, look, that data's there, have a play. I went off on holiday and he had a
play and I came back and he said, oh, something marvellous has happened. I
analysed that data, I released it into the public domain through this new
communication tool called Twitter. And loads and loads of biopetitions have all
said, yeah, they're interested and they've re-analysed the data, they look for
antimicrobial resistance, they've looked at how close it is to other strains.
It's like a crowd-sourced analysis. Wow, that's really interesting. I said, but
we're academics, we're not interested in our profile on Twitter. We need a peer-
reviewed publication on this. How are we going to turn it into a peer-reviewed
publication? I thought, well, we could write this up as a letter to the Lancet.
So I got some other biopetitions involved in it and we started writing it up as
a letter to the Lancet. Around the same time, I contacted the guy in Germany who
was heading up their efforts to analyse the outbreak and control the outbreak, a
guy called Martin Epfelbacher, and said, you know, we know a lot about E. coli.
We know a lot about genomes and bioinformatics, if you need any help, let us
know. And he came back and he said, no, I think we're all right. I thought,
okay, well, that's not an opportunity that's gone. I suppose we'll leave it. And
so I set up an opportunity with John Wayne, who was working at Public Health
England. I said, we can analyse one of the British isolates from this outbreak
and why don't we do it on the three different platforms available to us? So the
Iron 12, which we've got, the 454 Junior, which you've got, and we could
probably get a MySeq as well. And so we did that, and we did a little analysis,
and this is when Nick Loman came to his own. He went away and he did all the
analysis and he came back pretty much with a ready paper, publication-ready
paper, which we published. And we showed effectively that both the Iron 12 and
the 454 had this problem with homopolymers. Their accuracy wasn't quite good
enough, whereas the MySeq was much, much better. And that paper has now been
cited over 1,500 times, it's my second most highly cited publication, and the
Illumina reps used it as a way of flogging their MySeq for the years that were
to come. But while all that was kind of bubbling away, I suddenly got this phone
call at 10 o'clock one morning by Martin Eppelbacker saying, look, I want to
talk to you about this genome analysis and this paper we're trying to get put
together. Can I speak to you at midday? I said, you can speak to me now if you
like. He said, no, no, but we do need your help. And so I waited two hours until
he phoned me back at midday. And he said, oh, they'd sent their paper in to the
New England Journal of Medicine, and it had been OK in some aspects, but the
reviewers didn't like the way that the genome analysis had been done by their
colleagues at BGI, and they didn't like the way the paper had been written up.
And so could we help? And I said, well, I can help you write the paper. And I
got this guy, Nick Loman, who could redo the genome analysis. And he said, yeah,
there's a snag there. And I said, why? He said, it's Tuesday today. We have to
get this back to New England Journal by the end of Thursday. I said, I accept
your challenge. We analysed the genome, I rewrite the paper for you in less than
48 hours. I said to Nick, I've given you a challenge, Nick, go away and analyse
this genome and weave in all your stuff about the open source genomics as well.
And we did it. And so we wrote the paper and it went into the New England
Journal. I said, the condition of me doing this is I'm a senior author. And Nick
Loman is one of the first authors, joint first authors. And they said, yeah, all
right. And we got the paper accepted in the New England Journal. So that was one
of those things where we were in the right place. We'd lined up the things, you
know. I'd won the Iron Torrent. I had managed to recruit Nick Loman. We got Nick
Loman to analyse the Iron Torrent data. So we just lined up. So when that
opportunity came, we could seize it. And that's that saying of Chance Faving the
Prepared Mind. If you've got the wherewithal and an opportunity arises, you can
jump on it. I have a couple of comments. I mean, not really questions, because I
agree. It's an older story, but it's a keystone part.  of our history really
now. We couldn't have done like all this without it. I came into foodborne
infectious diseases in 2011 and this was like, I mean, there were a couple of
things going on, like there was a cantaloupe outbreak in the US, but like
immediately there was this E. coli outbreak. And then Nick posted on Twitter.
From my perspective, I just thought that was just weird. Like I did not know how
to handle Twitter. I think around then I went ahead and signed up for an
account, but didn't do anything with it. And it was just fun from the US side,
at least from my perspective, just to touch my toe in this and watch what was
happening. Yeah, yeah. No, it was a very exciting time. I mean, it was a
horrible time because it was an absolutely horrible outbreak and multiple
deaths. And there were deaths, you know, as we sometimes see with COVID and in
otherwise fit and healthy young people who just happened to be infected by a
pathogen. It was terrible. Yeah, I can't underscore that enough, but it was
awful. I feel like one other thing that maybe you didn't touch on as much was
how much the genomics played in figuring out what the pathotype of this thing
was because it was such a complex pathotype, right? Yeah. Yeah. So the
interesting thing was that we, going back to that statement, I said, seek
simplicity and distrust it. People working on type three secretion in E. coli on
the lean COVID system would say, well, it's always present in sugar toxin, sugar
toxigenic E. coli. It's part of that repertoire of virulence factors. It needs
to cause disease. And therefore it's very important. And we'd always say that
that's why we have to study it. And then this thing turned up. It didn't, people
have been kind of categorizing E. coli into different path of ours. And this
thing sort of straddled the boundaries. It was kind of sugar toxigenic. It had
some of the things that you have in interactive E. coli, but it didn't have a
lean coded type three secretion system. So it basically blew out the water, this
simplification that we've made of, oh yeah, this is part of it. So it was again,
a very illuminating thing that we always fall into this typological thinking
that we can nail things down by simple categories. And then it turns out nature
and evolution doesn't play by those rules. It just mixes and mashes and does all
sorts of things. So it's kind of exciting. So this was a lot of fun for me
because I was a doctoral student at the time working on anterohemorrhagic E.
coli. And my feeling after reading a lot of the literature leading up into this
was people's opinion. It had become incredibly dogmatic about pathotypes,
particularly for anterohemorrhagic E. coli. Paper after paper, just saying, you
know, if you didn't have the lead, we didn't care, you know, it was it was out
of the water. And it's a lesson that you have to learn and learn again. I mean,
around that time, I wrote a book called The Rough Guide to Evolution. And there
were all these quotes that people had said about Darwin, this and Darwin, that,
you know, Darwin and Marx and all that. And you scratch beneath the surface,
where is the evidence trail for that? And you find that actually, there isn't
one, it's not true. In fact, there was a quote that was attributed to Darwin
about it was not the fittest, but the organisms that are most adapted to the
environment that survived something along those lines that Darwin never said.
But there's a in the US, there's an august institution, I think it's on the west
coast somewhere that that had that engraved in marble, basically, when it was
shown Darwin never said that they look totally silly. So you know, you always
have to question the evidence trial. And it goes back to the early days of
genome sequencing, where when you say that this protein coding gene encodes a
protein that does this, where's your evidence trial. And often, you get this
very messy situation where someone would do a multi search of that protein, one
of the domains in that protein would have to something, and then that thing
that's had something that's got to hit something else, but they didn't actually
share a domain in common. And you just get this misannotation, all sorts of
problems like that. One thing I'll also add that was really exciting was as it
was rolling out on Twitter, you had rather than as a student, you normally saw
this, you normally saw results at this detail, at this level of of scientific
rigor being presented to peer reviewed papers. And then all of a sudden, you're
reading about the the entire aggregative like femoral genes on Kat Holtz blog
that she did with with with Nico Petty. And then you're reading the the tree on
someone else's website, then someone's tweeting stuff about comparison with all
of the existing I think David Studholm comparing it with all the other E. coli
existing E. coli genomes. It was a crazy time. Yeah, it did. It set the
precedent for what came afterwards in terms of outbreak control, open science,
you know, putting up pre prints rather than waiting until things been accepted
in the journal. And obviously, Nick and you and Andrew Page and many others have
adopted those approaches during the COVID pandemic. And I think that the event
that Nick kickstarted way back in 2011, it has borne fruit big time since then
in all these initiatives. I can't speak for Andrew and Nick, obviously. But for
me, this this outbreak reading about it sticks in the back of my mind when
approaching COVID. So a little while after we've had that New England Journal
paper come out, I said to the collaborators in Germany, who would never actually
met face to face, I said, look, we'll come over and see you. So we went over
there and we shared some champagne with them. And I said to them, what we got to
do though, show people we're not just a one trick pony here, what can we do now,
that convinces people that we've got something to say, and that we're cool guys.
And we said, what are our unique selling points? And I said, well, we're pretty
good at this kind of genome analysis stuff and bioinformatics, I suppose. And
they said, well, we got a freezer full of fecal samples. So I said, Oh, okay,
I've got an idea. Why don't we just sequence the fecal samples? Why don't we
sequence your fecal samples that you've got, and see if we can detect the
outbreak strain in those fecal samples? You know, it could be a kind of
diagnostic or clinical metagenomics. Why don't we try that? I said, it's a bit
of a long shot. And what I thought to myself was, well, if the dog were up until
that time, again, one of these dogmas was that E. coli was a bit player in the
in the gut microbiome, you know, less than a percent. So I said, you know, we're
not going to see, you know, it'll be the bacteroidetes and the Firmicutes making
up most of this. But we'll have a look. And if we even if we detect the sugar
toxin gene, that's still a hit, isn't it? You're achieving a diagnosis by
metagenomics. Nick Loman went away and did it, looked at it, came back and said,
I've got 20 fold coverage of the outbreak strain genome. In this one, I've got
50 fold coverage of the outbreak strain gene. And what turned out to have
happened against our expectations is, although in the normal gut microbiome, E.
coli is a bit player, during these episodes of infection, it undergoes a massive
bloom and takes over the microbiome, becomes the predominant organism. And so we
were able to get the whole genome of the outbreak strain from these analyses and
establish that you could do diagnostic metagenomics on the human gut microbiome.
And in fact, I met David Relman shortly afterwards, who was a key player in the
in the human gut microbiome project. And he said, I found it to you, we spent a
lot of time thinking about methods and protocols and getting it so that we can
analyse the healthy gut microbiome. But you've just come in here and said, why
don't we use this as an approach to diagnosing disease? And none of us have kind
of thought of that yet. So hats off to you. So, so that was that was cool. And
around the same time, I thought, because we'd had that success with
metagenomics, a guy came up to me who I'd worked with, was working with, called
Dave Minican. And he said, Oh, look, we do, we want to do some ancient DNA work
on TB. I've got some bison bones that go back over 10,000 years, I think it's
got TB in it. I said, Well, that sounds very precious sample. I said, I'd be
happy to have a play and do it by metagenomics, not mucking around with PCR and
stuff. But if we can detect by metagenomics, that'd be cool. And so I said, you
know, is there another sample that we could we could analyse that will be more
less precious? And he said, Oh, yes, I, we can get access to some, some, well,
some, some mummy samples, some samples from 200 year old mummies from Hungary.
And he put me in touch with a guy called Mark Spiegelman, who is a surgeon who
actually sampled these 200 year old mummies using a fibre optic endoscope and
took samples under sterile conditions. I said, Well, we'll take some of that
material and we'll just, I said to my postdoc, go and just extract some DNA and
sequence it. If we get a couple of reads of TB, that will be something, you
know, but I'm not expecting much out of this. She went away, she came back and
said, Well, we've got sevenfold coverage of the TB genome. What? 200 year old
sample and you've got complete genome sequencing? Yep. And so we went back and
we got more of these samples. And we ended up getting over a dozen of these 200
year old genomes out. And again, it was one of those moments where overturned
all your assumptions. Because the bioinformatician, Martin Sargent, who's doing
the analysis, he said, I was having trouble trying to work out what strain this
was, what's the closest strain. But then I realised that in the pipeline I was
using, it was saying that if a thing wasn't represented, positioned by a certain
70%, it was just masking that. But when I removed that mask, it turns out that
in many positions here, there's 50% of one thing and 50% of another. And so what
we've got is two strains mixed together, closely related strains, but two
strains mixed together. So what? You can't have TB twice, you can't get infected
with two strains. You only get it once.  you know, some cost over you, unlucky,
doesn't make much sense. But I went back and searched PubMed and it turned out
that actually, there was a literature that says you can have multiple strains of
TB. And it turned out that back then in the 18th century, the majority of the
patients we analyzed had more than one strain. And when you think about it and
you overturn your assumptions, if everyone was living in crowded accommodation
and everyone was coughing over each other all the time, why on earth would they
only get infected once? You know, that was very exciting. And around the same
time, an archeologist who I used to have a drink with in the bar in Birmingham
had been talking to me and I'd said to him, why don't we do some ancient DNA
analysis with you? And he finally came to me and he said, yeah, why don't we, I
want you to do some of this stuff. And I thought, oh, you'll give me a Pharaoh's
finger or something like that, something really exciting. I said, what have you
got? He said, I've got some mud, some sediments from beneath the sea, kind of
from a Neolithic transition. And I'd like you to look for sheep and goat DNA
because that would be the first sign of the Neolithic transition. If you can
find those in the environmental DNA, that would be really cool. And at the time
I was moving to Warwick and there I met a guy called Robin Allaby, who was a
plant microbiology, a plant genomes person, but he also done some work on
ancient DNA. And I said, look, I'm a bit busy. Do you want to take this on? He
took it on and he came back and he said, well, I haven't found the sheep and the
goat, but I found wheat. And wheat is as much a sign of the Neolithic transition
as sheep and goats in this context. So this is pretty cool. I said, wow, okay.
And he wrote it up and he managed to get it published in Science. So we got a
science paper on presence of wheat in the British Isles 8,000 years before
present, which is considerably earlier than everyone suspected. That's about
2000 years earlier than suspected. And that suggested that there were these
sophisticated social networks linking the Neolithic front in Southern Europe to
the Mesolithic peoples of Northern Europe. So it was kind of, that was pretty
cool. And then finally, in terms of metagenomics, I got into the chicken gut
microbiome around the same time, did a first analysis of that. And Martin
Sargent, who was working for me, he said, oh, I've done all this metagenomics. I
said, we'll do some of it by 16S, but we'll do a bit of shotgun. He did the
shotgun and he said, oh, I've done this kind of cool thing. I said, what's that?
He said, I've assembled genomes from this shotgun metagenomics data. And this
was before people came up with the term MAGS. This was back in 2012. He said,
yeah, I've kind of binned the reads using the tetranucleotide frequencies and
I've got all these genomes. I said, oh, that's pretty cool. So we wrote that up
and that became a very highly cited paper. And then to cut a long story short,
I'm still doing the same sort of stuff on the chicken gut today. And what has
happened is that we're discovering so many new species that I thought I've got
to name all these new species. It's not good enough just to call them bin 28.
And that took me into a whole other line of work of naming microbes, which then
led me into getting involved with the WHO to name the SARS-CoV variants of
concern. So I've been busy with these lines of work over recent years. We've had
a couple of other metagenomic kind of forays. We looked at the critically ill
patients in intensive care unit. And we found a similar thing to what we'd seen
in that E. coli outbreak that during these abnormal episodes, you get blooms of
pathogens. So we were seeing multidrug resistant pathogens coming up at 80 or
90% of the biomass in patients in the intensive care unit who'd received
multiple antibiotics. Which again, I couldn't believe it when I first heard it.
I thought, yeah, they'll increase a bit. They won't take over. But in some of
them, they were over 90%. You know, the whole microbiome is wiped out and one
organism takes over. So we found lots of interesting things through
metagenomics. So I just wanted to ask a final question about what is the future
directions then since you've seen so much change over time. So what have you
seen or read recently that reminds you of those early days as something new and
exciting in the same way microbiology was new and fresh? What's cutting edge
today or what do you see as the next foray for us as a field? Okay, so what I
have been surprised again, those moments of surprise are the things you live
for. When we started sequencing the chicken gut, we discovered 800 new species,
well over 600 completely new species and several that hadn't been named. And I
was just like, this is the most abundant vertebrate gut microbiome on the
planet. You know, more chickens than humans. Such a commonplace setting, but
we're seeing this remarkable microbial biodiversity undiscovered. And if you
run, typically we've done this with pig, we've done it with horse. If you take a
new context, samples from that new context and you run them through pipelines
like Kraken that compare them against what we already know, you end up with up
to 80% of the reads being unassigned. Some people have used the term func for
function unknown genes. I would argue that here, it's not just function unknown
genes, it's func in a different sense. They're phylogenetically unknown genes,
func with a PH because we can't even assign them to a taxon. We don't even know
where they come from. And it just shows us that we are really, there's this huge
ocean of discovery out there because we're up one or 2% in terms of the number
of microbial species that we have actually genome sequenced and characterized.
And so there is huge potential to discover new things. We will discover new
forms of life, we'll discover new branches. They're still discovering new phyla,
bacterial phyla. There is so much more to discover just through what you might
say is a brute force stamp collecting of just sequencing lots and lots of
things. And obviously we are approaching the stage where we can do the
sequencing definitively and properly, where we can get single contact genomes
out of metagenomes by using long reads and binning approaches. So the voyage has
really only just begun. I mean, we're 2% of the way there in terms of really
just mapping the genomes and understanding the genomes of the microbial world,
let alone understanding all their functions. So yeah, the future is very bright,
I think. All right. And on that note, I think we'll draw to a close. I want to
thank our guest, Professor Mark Palin for joining us today. We've been talking
about some of the bioinformatics highlights from his career, gleaning some
advice and thoughts for the rest of us. We've gone all the way from the start
all the right way through to the current day and a little bit into the future.
And so that's all the time we have for, and we'll see you next time on the
MicroBinfy podcast. 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 MicroBinfy. 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.