59: Hide & Seq: An ID Fellow Primer on Molecular Diagnostics

But really, I suspect that this will be helpful for anyone who is interacting with these tests because molecular diagnostics have been part of our ID space, but there are new developments every day, and some of us may not have the best handle on what these tests actually are, and more importantly, the strengths and limitations of these tests. Our guest co-host is here to guide us through helping you create a foundation for the terminology and methodology.

You may remember Dr. Pratik "Tik" Patel from our PJP episode number 53 with Josh Wolf. Definitely go back and check that out if you have not already. Tik is a second year pediatric ID fellow at Emory University and Children's Healthcare of Atlanta.

He also completed a pediatric hematology oncology fellowship at the same institution, and he wishes to leverage his training in both fields to advance the ID care of immunocompromised children with a focus on those undergoing treatment of cancer and hematopoietic stem cell transplant. He also has a research interest and introduction and implementation of novel diagnostics for improved stewardship and clinical care.

Dr. Patel's focused her research on improvement of next generation diagnostic techniques, understanding the inherent biology of periprosthetic infection, and understanding antibiotic resistance through a clinical lens. She has published over 500 publications and is the director of the Laboratory Center of the Antibacterial Resistance Leadership Group of the NIH. Welcome, Robin. Hi.

He is also interested in the process of selecting, implementing, and evaluating newly rapid diagnostic technologies using concepts of diagnostic and antimicrobial stewardship. He has received the Colorado Department of Public Health and Environment Astute Physician Award for work on acute flaccid myelitis and enterovirus D 68, as well as the Young Investigator Award from the Pediatric ID Society or PIDS.

He is currently the principal investigator of a multi-center pandemic preparedness clinical research study, PREMISE, pandemic response repository- microbial and immunologic surveillance and epidemiology, thanks for joining us.

And in those organisms we can use RNA as a target for detection.

You can really bring up the amount of Gen X a lot by pcr, and then you can detect that amplified material in a, a variety of different ways. Uh, traditionally, you know, we would run gels and do southern blots, but today we're commonly using probes. So, probes that are hybridizing at the same time that PCR is taking place so that we know we've amplified a very specific product which is important for diagnosis of infectious diseases.

But also there's what's called real time pcr, which is where the probes are hybridizing while you're doing your amplification. Um, and then you mentioned TMA (transcription mediated amplification) and there are actually other, what we call nucleic acid amplification tests or NAATs that are out there and they work on different principles, but the idea is there in that you're amplifying a specific gene.

So we see, um, lots of other sort of technical ways of amplifying and detecting a particular gene that are out there. And really, the term NAAT is a better term than PCR because it's, it's more general. Um, uh, but you know, technically some of the NAATs are pcr, so you can use that term in those scenarios.

And that signal will increase over time as you amplify your target. And at some point it will cross over whatever threshold it is that you are defining as a positive result. And so the cycle threshold means the number of cycles until you reach that threshold where you can say, Oh, I have a positive signal. And so a low cycle threshold means that you have a gene that you're targeting that's present in higher abundance than a high cycle threshold.

Now, uh, that's a general principle and in real time pcr, we often time have a cycle threshold value. In some of the other nucleic acid application technologies, we, we don't have a cycle threshold, but I think what people have been trying to do is to take cycle threshold values and transition what is a qualitative assay into a quantitative assay in infectious diseases. We do have quantitative assays, for example, when we measure HIV viral load, we have quantitative assays.

We're used to using those assays. Those assays are quantitative because they're specifically validated to be quantitative, and they include standards in the assay that tell you when you get a certain quantity, that's the correct quantity. They're not just a real time PCR assay. There's more in terms of quality control that's incorporated into those assays.

And so there has been some concern about relying on just cycle threshold values without creating a true quantitative assay, uh, around the COVID 19 pandemic. And sort of reading into the lab data too much. I think what people are truly doing is trying to make that test a test of infectiousness, which is a whole different conversation. But hopefully that helps answer your question.

So we've always had the ability, you know, since molecular diagnostics have come around to send a influenza PCR test that tells us yes or no is influenza there. And HSV PCR test that tells us is HSV there. Other examples, uh, would be Mycoplasma tests. There's now group A strep tests, many different examples of just targeting a specific pathogen and saying yes or no, is that present.

What we've seen is kind of a shift in the diagnostic approach from many of the clinical platforms that are coming out towards a more syndromic based testing. So we're combining multiple PCR test in one platform known as a syndromic panel. Um, and typically this is multiplex pcr, so semi nested PCR that can contain multiple targets, which can include viruses, bacteria, parasites, fungi. It can be DNA targets and RNA targets combined.

And basically, instead of saying you have to have a clinical suspicion of, is this hsv, yes or no? You can just say, I'm concerned my patient has a central nervous system infection. I'm gonna look for all the, are most likely suspects at the same time. So we can get into the, uh, benefits and the drawbacks of that approach, cuz it definitely comes with both.

Um, but you're seeing a significant shift kind of in the commercial platform field now that we have access to, uh, rapid identification of bloodstream infections. We have respiratory panels that can look for viruses and bacteria, both in the upper and lower respiratory tract. We have, uh, stool-based platforms that can look for viruses, bacteria, and parasites there.

Um, and we have the, the newer meningitis, encephalitis panels, each of which when they have been introduced, uh, into the clinical realm, have led to some twists of how do we interpret those, how do we use them? What clinical impact do they have? Are they cost effective? Um, and I think that's still a really interesting area of inquiry.

How do they change how we practice medicine day to day when we go from this era of having to have a clinical suspicion for what we're looking for versus kind of using a molecular platform to look for many things at.

Um, so going back to, you know, the basic root of diagnostic reasoning. How do I interpret that result in the context of the patient in front of me and, and knowing some specific caveats that detection of nucleic acid does not necessarily mean the presence of an active infection or the presence of an infection that's causing the symptoms that I'm evaluating for in the patient in front of me. So, Knowing that many viruses shed long after the active infectious period is done.

Uh, particularly common viruses like the rhino viruses, you frequently pick those up on the respiratory multiplex panels when they're not the cause of, of the disease in the patient in front of you, but they're just shedding virus from a, a previous infection. That's one aspect of it. Um, there is a potential for decreased sensitivity of some of the targets when you multiplex them.

So in particular, for example, the HSV PCR of CSF is pretty sensitive when you use a singleplex assay, that, uh, limit of detection is, uh, a bit higher when you use a multiplex assay. So you may miss low viral load infections in the central nervous system on a multiplex assay. Um, and then in general, I think interpreting a result that really just doesn't fit your patient context is important to take a step back and say, Yes, I detected this.

But is this truly what this patient looks like, smells like, sounds like is really important. And we've seen that time and time again with c diff. So detecting c diff shedding in stool when there's not the correct symptom complex in front of you. Um, as well as as many of the other targets like on the meningitis, encephalitis panel. Chromosomal integration of HHV6 is a huge problem.

So we see, um, about 1% positivity in the general population who just have the, the viral DNA incorporated into their chromosomes, and therefore you're gonna detect it in every cell in their body on, you know, anytime you test it. And so if you test, you know, like our lab 800 CSFs a year on the meningitis encephalitis panel, you're gonna have eight patients a year that have detection of HHV6, but that's not the cause of their disease. So it's hard to go through every example.

But just kind of going back to those roots of, of diagnostic reasoning and thinking, what was my pretest probability before I sent this test? And how do I interpret that in terms of the patient in front?

I mean, if you have different sort of versions of the same organism, you might need to make sure you capture that, but you actually don't wanna be capturing a lot of things because you don't want, you know, false positive results, if you will. But, um, there is another approach, and that is to go after a target that's present in every organism. And so the 16 S ribosomal RNA gene is a great example.

It's present in every bacterial species, and so you could use it as a general indicator that there are bacteria there. That is actually used in some diagnostics, but is not the most common way that this is used because there's another characteristic of the 16 S ribosomal RNA gene that's really helpful clinically, and that is so it's present in all bacteria.

It has areas of conservation and areas of variability, and so we can design PCR primers that target the conserved regions that will amplify a fragment of the 16S ribosomal RNA gene from any bacterium that's present in the sample. But then we can sequence the area in between those primers and sequence through variable regions that tell us based on looking at that sequence data, which bacterial species it is that we're looking at, and that gives you what we call a broad range bacterial approach.

There are other genes that are conserved in bacteria that could be targeted. But we have the most information of any gene for the 16 S ribosomal RNA gene. So that's the one that you're most likely to see, um, in an assay. The sequencing itself is something that has been done routinely in clinical laboratories like ours since the 1990s, but that has used Sanger sequencing.

Sanger sequencing essentially lets you interpret a very clean sequence read from a single bacterial species that has no copy variance of its 16S ribosomal RNA gene because oftentimes this is a multi copy gene in bacteria.

When you have more than one sequence of the 16S ribosomal R RNA gene present in a sample, such as in the case of a polymicrobial infection, if you attempt to do Sanger sequencing, it's almost like reading two words at the same time with the letters on top of 'em, and you cannot easily decipher what it is you're looking at, but that can be sorted out now with next generation sequencing of that product.

And then we can, we can really look at the full portfolio of anything contributing to that 16 s ribosomal RNA gene sequence data. It's actually a technique that's used to define the microbiome in microbiome research, but it can be used clinically on samples that don't typically have a normal microbiome to, um, sort out when, uh, there's either a very low amount of bacteria present in the context of some background or more than one bacterial species in the same sample.

This gene is, uh, specific for bacteria so it doesn't pick up other organism types like fungi or parasites or virus.

And then Kevin had talked about multiplexing because we see a lot of these panels that are out there today that we use that can do multiple PCR at the same time. So theoretically you can do all of that, uh, together.

And some areas are more informative than others in separating out species of different groups of bacteria such as Mycobacterium species. But certainly the 16 s ribosomal RNA gene can be used, but then you can look at other targets that might be perhaps more informative in Mycobacterium species to maybe get a higher level of, uh, separation. I think diagnostically. Uh, there are two questions. You know, if you're thinking mycobacteria, one question is, is there a mycobacterium present?

And another question is, what is that species of mycobacteria. You know, sometimes, we can't get to the detailed species with, with many of these diagnostics, even Mycobacterium tuberculosis is a complex, uh, but many others are groups or complexes of organisms as well, which is probably fine clinically for the, for the most part. Uh, but again, I think the main message is that mycobacteria are bacteria.

. Pratik Patel: And when you talk about sequencing, can you speak to how does the identification happen specifically? Like what, what is done with the sequencing data and then how does that match up to like figuring out which organism is causing the or is present. Yeah, that's, uh, that's the fun of sequencing. So you generate a lot of data. And I'll talk first about Sanger sequencing because that's the most straightforward.

You know, you have a string of nucleotides that comes off, typically, um, you're doing bidirectional sequencing because you're sequencing from both the forward and reverse primers. So that's nice because then you can overlap those and you know, you got the same answer twice in both directions. So it's a, a measure of quality, I guess, that, that you get. Um, so then you take that, um, concatenated sequence that's been put together and you have to run it against a database.

And, and this is where, um, there can definitely be some variability. So either you're using a pre-constructed database where someone, either your team or others have put together a database that says, you know, this sequence is Mycobacterium tuberculosis complex, and now this sequence is Streptococcus agalactiae et cetera. Or you're using a public database such as NCBI, um, that database is going to have a lot of sequences in it. It's not completely curated, uh, but it's more comprehensive.

Um, and so your analysis has to really look at what does my query against this database tell me this is, um, is it tell, And it could tell you. That it's this species, and then you have to determine whether it's all the related species to that species have been considered in your analysis, are in your database, and that there's maybe enough distance from anything else to be able to call that particular species. Uh, sometimes they're not.

There's not, and there are a lot of organisms that read in together. Here's an example of that. Brucella species, they all pretty much have the same sequence. Actually, they're probably all the same species, and that'll happen probably sometime in your future ID fellows just to make things confusing. But, but so, you know, you couldn't possibly call a particular species of Brucella based on that analysis alone. But we know that, uh, clinicians like to know as much information as possible.

So you try to get to the species where you can get to the species and otherwise you group, kind of, um, roll up to a genus level identification or a group or complex level identification, or most closely related to if there's something else reading in that's, um, maybe, you know, nipping at your, your heels, um, behind that sequence.

Um, next generation sequencing is more complicated because there, if you're sequencing bidirectionally, you have forward and reverse sequences, but you don't necessarily know what goes with what. And oftentimes you have multiple different organisms that are reading in together at various abundances. And so interpretation of that is done in the same way against databases, uh, but, but is a lot more complex.

Um, a lot of times, especially if you're looking at a clinical specimen, that doesn't have a lot of organism in it, you are also seeing, um, the background sequences of the assay because there are bacteria everywhere. They're all over us. They're in the environment. They're, they can be in reagents. And so if you, um, if you dig deep enough, you'll find bacterial sequences, 16S ribosomal RNA, gene sequences in pretty much everything.

So then you have to sort out, not only what is it, You know, is this something that should be clinically reported? Because we, we all know how much confusion that can create. Um, Kevin spoke to that a little bit, even with the multiplex PCRs. Uh, but you know, when you report out something that maybe is just coming from your background, but um, you know, when it comes out in the report, looks like it could be clinically significant. So, um, databases are what, um, what you need.

I will say another interesting and unique challenge here is that, uh, bacteria, bacterial taxonomy is rapidly changing. So exactly what you call things can change. And, um, in our experience we've also seen sequences of bacteria that probably are not yet named. That's a real challenge to report on the clinical side. So there's a lot of work that needs to be done to continue to describe bacterial species.

That that is not related to that and doesn't belong in this genus or, you know, et cetera. And so then they get around to renaming things and, and then we have to update systems and change the way we report things. And then that causes a lot of confusion on the clinical side and, you know, undoing of what's been taught in the past and so forth. So nobody, nobody loves those changes. . Pratik Patel: Yeah. And who does broad range PCR in the US?

Can you guys speak, each of you speak to some clinical scenarios where there's data or, you know, personal experience of yours, that instances it could be useful for ID clinicians.

There's a broad range of experiences with the use of them and I, I would, from a clinical research standpoint, just caution folks, when you're reading retrospective observational data sets, just know that that tends to be very heterogeneous population and oftentimes sent at various time points, often later in the course of disease, things like that. So there's a lot of caveats to that data.

Um, I will say having read a lot of the observational experiences that the yield is lower than I would expect for many of them. Whether that's due to, you know, patients being pretreated or being sent late, um, in our hands, from a personal view standpoint, the 16S, 28S kind of platforms have been most useful in situations where you have source tissue, so you have a biopsy.

You see organisms, so it's Gram stain positive or you see fungi there, and for whatever reason, whether it's a diagnostically challenging organism to grow or a situation with pretreatment, you can't get an identification by routine, uh, microbiological techniques. Those tend to be the situations in which we send those out, and those tend to be our highest yield situations

The most recent one is in clinical infectious diseases this year, and we did a look back our 16S ribosomal RNA gene PCR sequencing assay applied to 2,146 normally sterile tissue and body fluid samples in our routine clinical practice. Um, we do an algorithm where we run a PCR assay first, and um, we get a CT value from that PCR assay. And if the CT value is low, we run Sanger sequencing.

If it's medium high, then we go to next generation sequencing because our validation data tells us that Sanger sequencing is unlikely to give us a definitive answer there. And if the CT value is high, we, we know that sequencing by and large doesn't give us a useful result, so we just report that result as negative. It's a relatively parsimonious way of applying this kind of testing.

And, um, so what we found is that adding this next generation sequencing to um, just the Sanger sequencing approach increased our positivity rate by 87%. Um, and you're right, uh, Kevin, that maybe detection rate is the best way of looking at it is, is not maybe as high as you would love to see it, but I, I think you're also right that there are certain scenarios where this kind of testing can be particularly helpful.

Obviously when you're suspecting a bacterial infection, but when you can see the organism on staining or there's a histopathologic response that suggests there might be a bacterium there, that can be really helpful. And um, one disease in particular that I'd like to highlight, where I think this is really standard of care is in infective endocarditis, but in a particular scenario. Um, we all know that blood cultures are the first microbiologic tests that you would do in that scenario.

And of course, if you get positive blood cultures with a consistent organism, you don't really need to do other testing. And then if cultures are negative, which sometimes they are, oftentimes because of antibiotics that were given prior to blood cultures being collected, um, Then you go on with your culture negative endocarditis workup, uh, with a Brucella serology and a Coxiella burnetti serology.

But we don't have great diagnostic tests, um, yet for some of the other common causes of infective endocarditis in culture negative cases. And so if patients do come to valve resection, um, that valve should be sent to histopathology for one thing. Uh, for an, an expert histopathologist, someone who has experience in infectious diseases and, um, cardiovascular pathology to take a look at that valve.

And, um, if there's acute inflammation there, if it looks like it's consistent with infective endocarditis, then running a 16S ribosomal rna uh, gene PCR sequencing assay is really your test of choice, actually above and beyond culture. Um, we've also found some, some interesting detections in plural fluid, which I think is a clinical type of specimen that we're still learning a lot about as far.

Um, you know, what's going on and what microbes are causing pathology in people with pleural effusion. So that can be very helpful as well. Um, But again, we continue to learn from this. Um, we know that antibiotics affect the sensitivity of culture and they also affect the sensitivity of 16 s ribosomal RNA gene pcr and sequencing, although to a lesser extent, uh, but not surprisingly, since they target bacteria.

Um, Detection rate goes down in people on antibiotics and it goes down progressively, um, uh, depending on the amount and timing of antibiotics that have been received by the patient.

And if you do that, starting from DNA, you'll sequence the human genome. You'll sequence bacteria, fungi, parasites and DNA viruses. You'll miss RNA viruses, of course. As with PCR, you can introduce a step where you take RNA and you convert it to DNA and sequence that, and then you can detect RNA viruses as well.

But you can also do next generation sequencing on amplified single genes like the 16 s ribosomal RNA gene that we just talked about, which is more of a targeted metagenomic approach as opposed to, uh, a shotgun metogenomic approach, which is used when you're sequencing everything.

Um, I think the most, uh, pertinent one that we're seeing a lot emerging on the clinical side of things, um, is the plasma cell-free DNA sequencing technology. So the commercial test is called Karius currently. Um, so you send a, a plasma specimen out. It gets, uh, sequenced for trace amounts of microbial DNA and plasma. Um, it can detect both organisms in the bloodstream and trace amounts of, uh, DNA based organisms.

Uh, even in tissues in some cases, we're still learning more about the sensitivity and, and specificity of various targets on that platform. Um, but we're seeing more and more of its use for clinical care in diagnostically challenging cases. Um, another is, uh, CSF uh, next gen sequencing for meningitis and encephalitis. This is a category where we've never been able to chip away at that 50% of suspected cases of CNS infection that we just can't get a microbial diagnosis.

That can be caused by RNA viruses, DNA viruses, bacteria, fungi, parasites. So it's kind of the perfect application of a very common clinical presentation that's very hard to differentiate etiology based on clinical factors, um, and could be due to many different things. So there is a, a clear approved platform that, uh, is being used at, uh, U C S F, uh, Charles Chiu Lab, um, that can do sequencing of CSF for those diagnostically challenging cases.

Unlike the Karius that we talked about that just detects DNA based organisms, this detects DNA and RNA based, uh, organism. Um, a little longer turnaround time, typically with that platform. And then we talked previously about the broad ranged, uh, and, and sequencing platforms used for, uh, tissue samples from source, uh, samples, biopsies, and others. So those are the main, uh, platforms currently clinically available.

I will say that technology sometimes advances quicker than our knowledge on how best to use them and how to interpret them. So I think we're in the catching up phase of we've got these really neat new tools. We're learning how to, to best use them and how to best apply them for clinical care.

As an example, um, my team works a lot on periprosthetic joint infection, which is largely a bacterial infection, and we have a publication in press Clinical Infectious Diseases, where we compared, uh, 16 s ribosomal R gene PCR sequencing or targeted metogenomic sequencing based approach to shotgun metagenomic sequencing on a specimen we call sonication fluid, which comes from removing biofilms from explanted devices, and, um, the performance of the two approaches was the same.

And that makes sense because you're just looking for bacteria. But I think it's also really helpful because it's a lot simpler to do a targeted approach and a lot more cost effective than a shotgun approach. Um, and so in clinical practice, I think we need to sort out disease by disease, when to use these tests and exactly which test, uh, to use. Uh, and so there's a lot of research for people interested in research in infectious diseases.

A lot of very clinically relevant research that lies ahead.

So not just sending it at whatever time point for whatever disease process, but how does this particular platform work for culture negative endocarditis like Robin talked about, or prosthetic joint infection, um, or MSK infection in pediatrics. Cause I think that's gonna really inform when we should be using this test upfront. So not sending it after a week of, you know, getting. Getting no diagnosis from conventional diagnostics and when is it really clinically impactful and cost effective?

And we are just not there yet with the data that we have. And in a way, it's kind of backwards to the way in which we do drug development, which you go after an indication, you show clinical impact of that drug for that indication. It's like we're doing that all backwards. We have the new tool and the new technology. It's approved because it works in the lab. It can detect what it's supposed to. But then we're trying to figure out what use it should have.

And I think there's a lot of research work on the clinical research side, as Robin mentioned. Um, That's ripe for the picking of ID fellows and others interested in diagnostic stewardship to do that kind of backwards work of how do we take these new shiny tools and apply them to their best.

And, you know, we may do very sensitive plasma cell-free DNA based sequencing and detect all the things in their gut that are spilling over with trace amounts into the bloodstream. Likewise, in CSF, uh, there can be interference from host background as Robin talked about. When you do, uh, metagenomic next gen sequencing, you not only get the pathogenic sequences, but you get all the host background sequences in there.

And if you have that tap that's either bloody or has a bunch of white blood cells that can actually interfere with the ability to detect organisms there so often you'll get uninterpretable result as far as that goes. And then it's the great unknown too, so you can sequence pathogens that have never been described before, that have never been described in that disease process. Um, that's one of the fun and interesting and challenging parts of this.

So we were part of the initial study of the next gen platform for CSF that is being used at U C S F. And we used to have a weekly, we called it a tumor board, where we would go through all of the positives from the week prior from these really interesting encephalitis cases and try to interpret what had been detected on sequencing and what do you do when you have a patient with severe encephalitis and you detect a virus that's only been described in crickets?

Is it truly a cause that we've just never found before or is it you're just detecting uh, some lab contaminant or something else? So there's a lot of clinical interpretation that needs to still occur with these assays.

There's a lot of caution we need to take, but I do think every diagnostic test has its place and I think we've all had a case in which, you know, we haven't been able to get to the bottom of it, and we send a sequencing assay and all of a sudden it becomes clear like there's something that's been there that either we didn't think about, or it's so rare that it wouldn't have been on our radar and really improves the care of that patient.

So I'm one who believes that every diagnostic test has its place. We just gotta figure out the best way to use it in the best situations and make it work for us the best of our ability.

I hope maybe in the future there's a way of sharing some background clinical data, um, on the patient. I agree with Kevin. It's, it's nice to know what people are looking for, but you really wanna understand, you know, is what I am seeing, in any way, shape or form, possibly fitting with what you're seeing because sometimes you don't know what you're looking for exactly when you're running these assays. So I think we're learning a lot.

Um, we've discovered new organisms and new diseases along the way, and that will probably continue to happen. So it's really exciting. I, I definitely think these assays play a role for some patients, and we have to figure out who those patients are. And many times they're today being run as a test of last resort, which is okay because we're learning, but probably during the career of some of the ID fellows, these will become like a first line test so we can get the results back more rapidly.

And I think you'll see improvements, I hope, in sequencing, uh, technology that will drive costs down because some of these tests are still very expensive. Um, and you know, that's a problem in healthcare. And then, results are not necessarily rapid, not rapid in the way, you know, some of the multiplex panels are delivering one hour, uh, results. So, um, you know, look, look forward to learning more and also seeing technology and, um, clinical understanding improvements over time.

And. Lots of institutions who've decided to roll out these tests have rolled them out with some hands on approach of requiring approval, either by the ID team and an ID consult or the clinical micro lab director to review the case ahead of time. So you don't get stuck in that situation either where you're sending it in a very low utility situation and wasting a whole lot of money or interpreting a test result that you never really should've sent that test in the first place.

So I think thinking about from the micro lab and clinical ID standpoint, how to roll out these tests and potentially putting in a few more handholding, uh, ways to kind of guide their best use is probably the right way to do it, at least until we get better data to drive their use. Like Robin said, my goal hopefully is someday we have great data that says, for this indication, send it the day that patient comes in and we're gonna improve their care.

But until we get there, there probably needs to be a little bit more infectious disease or, or lab involvement in the use of these tests.

And if we know how to analyze that data and go from genomic data to phenotype susceptibility prediction, then we should be able to get all the way there. This is very futuristic, right? And it's probably going to be contingent on having enough organism in the sample to actually interrogate the genome in that way. But I believe it will be possible in the future to get susceptibility information as well, which hopefully people are excited about.

And one of the interesting part of, uh, Metogenomic next Gen sequencing is not only does it generate, you know, sequences that we can look for pathogen in, but also sequences out the RNA transcriptome. And so if you categorize the genes that are being turned on in the host, there's actually some machine learning work to say that response looks like an autoimmune response or a viral pathogen host response or a bacterial pathogen host response.

And that might be enough to at least send us down the right pathway in those patients that we can't get an organism, but we need to know whether, you know, the neurologist should immunomodulate that patient or we should stick on empiric therapy cuz we might not have grown or detected an organism.

So I think we might be actually stepping back from the precision diagnostics of getting a specific name for something in those patients and just saying, uh, better part of valor maybe is at least put them into a bucket of autoimmunity versus infection versus other.

Stay safe and I'll see you next time.