 Welcome everybody. The recording is in progress. You're all being, I'm being recorded. Welcome to day two. We've got some really exciting talks. Three talks, and then there's a discussion. The speaker, Alejandra Guter, I don't know how to pronounce the surname. Okay. Unfortunately cannot make it. So we were planning to have a block of time where everyone who signed in gets to introduce themselves and talk briefly about what they do and whether there's anyone else present at the meeting who they'd like to meet. There are two types of things that happen, I think when I'm in a conference. One is that there's someone who I think is really interesting to meet, but I haven't got anything concrete to say. And so it's a bit embarrassing, but I want to meet them. And the other thing is there's someone specific who I want to meet and ask a specific question. So, I think in that break we're going to go around to everyone who's willing to do it, just to introduce themselves and if there is someone else who they want to meet, either to say something specific or just because they'd like to meet them, then you can announce it then and then people can organize individual meetings as and when they like. Anyway, can I say one thing? Yeah. Matthew suggested that since some people may not want to talk to say anything, the people who want to talk just leave the camera on. And that moment not now. So that we know when we don't bother people who don't want. Yeah, that sounds good. That way we won't chase people. Okay. Let's start with the first talk. This is Will Matlock from USF Oxford. Talking about plasma distributions in human bloodstream infection associated and non-human entrobacterial is in Oxfordshire UK demonstrating sharing across reservoirs. Will, are you online? I am. Yes. Thank you so much. Brilliant. Oh, there you are. You are sharing screen already. Yes. Thank you so much for having me. Let me minimize everyone's faces. There we go. Yes, thank you so much for having me. Today, I want to give an overview of a current study in progress, which is looking at the overlap between human and non-human entrobacterialis plasmids. And I really hope this talk gives some insight into One Health for AMR. We have Tim Walsh's figure there. One Health for AMR from a plasma perspective. Let's begin. Our ambition was to try and unpick the dynamics of plasmid sharing across human and non-human compartments. Plasmids, as was discussed in depth yesterday, often carry clinically important antibiotic resistance genes. So understanding that their evolutionary histories is crucial to managing resistance as a whole. The recent studies have started to unpick plasma sharing, but often are limited in size given the genetic diversity in these niches. Restricted to single species or phenotypes, for instance, drug resistant isolates, and have not fully evaluated the dissemination of MGE, such as plasmids as we're talking about here or insertion sequences. And this is often due to fragmented genome assemblies. And there was also some discussion yesterday on assembly techniques for plasmids. So to address previous limitations and better explore entrobacterialis plasmid diversity and sharing across compartments, we assembled both a geographically and temporally restricted sample to investigate two avenues, which I want to talk about today. The first of which is how our identical plasmids share between compartments. And the second component is how is broader plasmid evolution history playing out between compartments. And the main emphasis here is on human and non-human sampling, which is of course of clinical interest. So the data set itself comprises of large isolate collections from human bloodstream infections from 08 to 18, and from longitudinal sampling from the rehab project, which includes livestock, environmental soils, wastewater, and the waterways kind of surrounding wastewater treatment works. And this totaled quite a hefty data set of around one and a half thousand complete entrobacterialis genomes, meaning both the chromosome and all the plasmids are circularized. And indeed that included over three and a half thousand circularized plasmids from this restricted data set. And briefly, the genomes were unicycler hybrid assemblies. So we were using a lumina, pet and short reads, and then either pack bio or nano for long reads. So ideally this will capture both the global syntony gene ordering of the plasmids alongside a high quality nucleotide resolution, which of course you need for things like determining ARG alleles and so on. So to begin with the most simple approach. This was pulling sets, outsets of plasmids from the data set within a very conservative, conservative mash threshold and a generous sketch size. And this revealed this plot at the top revealed that most compartmental overlap of neuroidentical plasmids came from the smaller, often well conserved colotypes. So we see this axis here is increasing in length and towards the left hand side we have these compartmental breakdowns of the various identical plasmids, blue, the BSI is green, the environment's red, the livestock's and purple, a combination of waterways and wastewater. And then as we move rightward, we often see longer, often resistant and conjugative plasmid sets, which what these bars below indicate. But we also see more segregation between human and non-human compartments. So perhaps greatest interest here are these two conjugative F type plasmids, this little bar below the red arrow, carrying multi drug resistance efflux pumps, both in E. coli ST54. And they were found in a cattle and a BSI isolate. And I'll come back to these later. But overall the general message is there is a much overlap of identical plasmids. And of course, again, as was discussed in depth yesterday, plasma devolution is far from vertical and take together the action of other MGE's recombination, co-integration and other such events. Plasmids evolve in highly complex dynamics. And a lot of these changes take place within a well conserved plasma backbone. I have a quote there from Alex Olex Brilliant 2018 paper. What we often see within plasmids is sets of core genes with this well conserved syntony within which accessory genes come and go. So what we really need to do is look beyond identical sharing to understand how compartmentalized or segregated plasma devolution truly is within our sample. So we're interested to kind of briefly summarize our methodology for the second step. We began by clustering the plasmids by the Jack and Jacquard index if there came a content 21 more content using mash with the generous sketch size of a million. And then we looked at the plasmids with the Levain algorithm, which I believe was also mentioned yesterday within a plasmid plasmid network. We then annotate star plasmids with with prokka which uses protocol and abracad with various databases including including plasmid finder. And where this is all going to end up is examining the plasmid cluster core gene phylogenies generated with panoroo to pull out our core genes and then IQ tree to generate the phylogenies. So the actualized here is the quite an unwieldy looking network, but this this is our weighted jack on similarity plasmid network threshold of the similarity of point five. So what this means in practice is any two plasmids within this network are connected, share at least 50% of their mutual 21 months. And briefly, because I know we have a lot of plasmid networking type people here. That threshold was chosen by examining the component evolution of the network as we stripped away edges. And again, colored within the same compartmental color scheme, we immediately see some cross compartmental genetic sharing. And that's really what we wanted to explore further. And so as I mentioned, we then cluster the data set we segregated our plasmids into into 247 clusters of at least three plasmids using the Levain algorithm and they're shown here in varying colors. And by doing this we found around 30% of our clusters contained both human RBSI and non human livestock or environmental plasmids. And as an aside here, in terms of antimicrobial resistance the folks of this week I over exaggerate them for super clarity here in red. In terms of clusters we found plasmids carrying antimicrobial resistance genes in 21% of our clusters over 550 plasmids. And it's also worth noting here that AMR carrying plasmids were present in about two thirds of AMR carrying cluster plasmids. And this really highlights that the AMR genes are not necessarily widespread on genetically similar plasmids potentially having to transpose on action. And this is why potentially plasmid studies should ideally consider the entire plasmid dome of the niche in question, not just the resistant plasmids or resistant isolates. But this figure here at the top summarizes from the previous slide that the 69 10 plus member clusters decreasing in size. What we see is again this kind of range of compartmental representation. But we also see they rent represent a range of bacterial hosts of mobility's lens replicant haplotypes and of course resistant profiles here colored by beta lactamase and non beta lactamase carriage. And these clusters tend to have have high purity with respect to these metadata is the often single color bars below these clusters represent. But what we really wanted to do was examine the genomic structure of these plasmid clusters. And so as I mentioned a minute ago, we did a Pangea name style analysis of these clusters using panoroo. And this pulled out a set of core genes from each cluster. So these are genes which almost all plasmids shared. They include proteins of replication mobility establishment, effectively the core functionality necessary for a plasmid within their niche. And though importantly is pretty actually important to mention that, though often these genes do align with with with the functions we would expect they were not restricted by functional rotation. They were rather just like nostically chosen at a 95% threshold. The remainder, the accessory genes are often niche adaptations the antimicrobial resistance genes we saw above, virulence traits heavy metal cassettes toxin anti toxin systems and so on. And they have added flavor within this, this putative plasmid backbone. And the bottom panel here shows us somewhat the number of core and accessory genes per plasma cluster in orange and blue respectively. The core genes on average comprised a median of around 40 odd percent of the total cluster Pangea name size and thankfully Pangea name size was also significantly correlated with plasma lens. It's a very open Pangea known by the looks of things. So then, again, as I mentioned earlier we we produce phylogeny to these concatenated core genes. So here's an example. This is a core gene phylogeny for clusters 35. This is a group of conjugative beta lactamase producing I to type plasmids found across human cattle pig and sheep compartments. The right to the left was produced in IQ tree with alignments by Maft by concatenating the the 50 core genes in the cluster. That's those found again within within 95% of the plasmids, which for a cluster this small is is, in effect, just 100%. And then the heat map to the right represents an average ordering of the cluster gene repertoire. So we've colored them by core accessory or transposes. This really shows that the putative persistent backbone structure within the cluster upon which accessory genes are gained or lost. If we now color the tips by E. coli sequence type, we also see that these these interrelated large plasmids are found on a variety of lineages. So this plasma does not linked also to a to a clonal isolate. Also, and especially looking at other pan genome heat map floor genes yesterday, you might notice that the the the co phonetic distance of the plasmid core genes, alongside the the the accessory patterns to the right we can often see concordance that's to say that clades seem to kind of match up with the heat map. And we plotted this relationship. So here we have along the x axis, the core gene co phonetic distance and on the y axis, we have the jack our distance of accessory gene presence in absence. And what we see is that the low core gene distance and high accessory gene distance is more common than the high core gene distance and the low accessory distance is kind of nether zone to the bottom right. So just the interpretation here is is how the plasmid backbone evolution is is our slow our slow march, and then the accessory gene gain or loss is kind of our fast, fast signal. And this may give some clues as to as to how plasmid evolution could be modeled. Yesterday we started with Amin mentioning kind of traditional file genetic models. And then towards the end of the day, Liam was discussing structural discrete genetic gains losses and rearrangements. So perhaps any model of a plasmid evolution should reconcile this this slow march of the plasmid backbones, alongside the discrete changes in accessory function. And of this pattern we indeed found the same for other clusters here with at least 50 accessory genes. And again, we see these kind of nether zones to the bottom right this this trend persisting of this slow core evolution versus this fast gene accessory change. So maybe I'd like to finish with the more with a more concrete example instead of these kind of high level overviews. So here is another example within cluster 10. And this is a group of resistant again, conjugative F type plasmids, many of which are carrying multi drug resistant efflux pumps. I just want to draw attention to a simpler example of plasmid evolution of stable backbone accessory movement across human and non human compartments, specifically revisiting that example way back from the start of our near identical BSI and cattle plasmids. With the addition of a soil plasmid from near a poultry farm. So this here is an alignment of those three plasmids using clinker and the top two cattle and BSI plasmids are about five snips apart, but the poultry plasmid at the bottom. If this will move again. There we go. And it features the relocation of this insertion sequence of this isk PN 37 downstream and also features might take a couple seconds. It's quite a long plasmid. It's about 105,000 base pairs. It also features the addition of this transposes and this this hypothetical protein. This is a very, very simple example of how rearrangements and genes gains and losses from a main path of plasmid evolutionary history, and also shows how similar plasmids appear to be evolving between human and non human compartments. And so to bring kind of everything together. Our data doesn't capture much sharing of near identical plasmids between human and non human compartments and really what we do have a just well conserved quite prolific cold types. But what it does give is a kind of a body of evidence for more intertwined complicated evolution where similar plasmids differ by a handful of accessory genes. I would also like to emphasize a point I made earlier and that's that resistance plasmids are often part of a far broader plasmid dome of a very similar but non resistant plasmids. And as such it makes, to me anyway much more biological and ecological sense to study the whole population when possible. Our next steps with the study to try and date these events. We've had a had a go. A very limited go with with using IQ tree with preliminary findings showing some divergences of human and non human resistant plasmids within recent years or decades, but this is very much a methodology which needs refining. I'd say thank you to my supervisors, Sarah, I think Nicole and Liam are here today. Dan Reed from the CH Derek Crook, other people in my lab, Sam Bede, and of course everyone from the APHA who allowed us to sample their farms and Moon and Monal, Richard, everyone at APHA. Thank you very much. Fantastic. Well thanks very much. That was a great start of the day. Do we have any questions from anyone up front? So I did raise my hand. Oh, sorry. Maybe you don't see me. I didn't go ahead. I really enjoyed your talk a lot. I was just curious in your environment, the collection of isolates. Do we also have isolates this plasmids coming from produce or vegetables. So this is effectively an amalgamation of two separate studies so BSI study led by Sam Lipworth and others in my lab, and then the rehab project which is kind of spearheaded by Nicole and others vegetables were unfortunately out of the scope. You will hear in my talk, because I'm reporting about E. Coli carrying plasmids coming from lettuce and cilantro and so on. And it was quite interesting to see that we have quite a high diversity of these type of plasmids in E. Coli from cilantro, for instance, and obviously this could be a very nice link from the environment to humans via the gut microbiome. Definitely it's all about trying to kind of postulate on those links within the One Health Network. Thanks very much, Cornelia. I'm going to quickly ask one question that's sitting in the in the chat will from Sonya Leighton and she said thanks for the great talk will you touched on this a little at the end a little bit but I was wondering whether you could say more about what slow and fast mean in the context of the evolution is it possible to put some estimates on what these rates are. As for rates for the for the core genes, we can extrapolate the number of substitutions per base from the ML trade. As for the kind of the accessory genes which are moving through generally through transposons and stuff that's probably far, far harder to characterize and then probably more of a job for the experimentalists or for far, far larger data sets. But what we do certainly see is is different insertion elements moving between different kind of fixed backbones within the data set. Okay, I think that question of rates is going to come up again and again this week. And we have another question from Ambrine Couser but we're already five minutes overdue. So, I'm sorry, I'm bringing we're going to save that for the discussion later, but we'll feel free to answer on the chat.