Efficiently Following the AMP Guidelines with VSClinical and Golden Helix CancerKB

About this webinar

Recorded On: Wednesday, September 11, 2019

Interpreting somatic variants for the clinical genetic testing of tumors requires hands-on time of the most skilled clinical lab personnel. Various clinical and genomic sources must be queried, papers and guidelines referenced, and an evaluation of the clinical actionability of the mutation determined by the following the AMP guidelines. Yet, there is tremendous potential for reuse of this time consuming and valuable work!

Earlier this summer we launched our VSClinical AMP Guidelines workflow which integrates a lab-specific knowledgebase that saves every biomarkers interpretation in up to seven different snippets that are reusable across various genomic and clinical contexts. Furthermore, our cancer workflow is bundled with the Golden Helix CancerKB, our expert-curated interpretations of the most common biomarkers for the most common cancer types, reducing the time to your first precision medicine report.

Join us as we cover:

  • The interpretation of clinically actionable biomarkers for targeted molecular therapy and diagnostic/prognostic clinical reports for cancer
  • The different levels and scopes of re-use of the interpretation for each biomarker
  • Saving, re-using and updating these interpretations over time by multiple users within a clinical lab with an integrated lab-specific knowledgebase
  • The built in Golden Helix CancerKB that provides default interpretations for most cancer genes, biomarkers and many clinically actionable Tier I/II drug sensitivity and resistance interpretations.

In the world of genomics shaping precision medicine in oncology, the limiting factor is the time-to-sign-off on fully interpreted molecular profile reports.

Watch on demand

Please enjoy this webcast recording. Should you have any questions about the content covered, please reach out to our team here.

Download the slide deck

To download a copy of the slides, click on the LinkedIn icon. This will redirect you to the SlideShare site. From there, you can clip your favorite slides or download the entire deck to your computer.

Love this webcast? Check out more!

Find out how Golden Helix software enables users to harness the full potential of genomics to identify the cause of disease, improve the efficacy and safety of drugs, develop genomic diagnostics, and advance the quest for personalized medicine.


Hello everyone and welcome to today's webcast. Thank you all for joining, my name is Delaina Hawkins, Director of Marketing here at Golden Helix and it is my pleasure to introduce today's presenter Gabe Rudy, who is our VP of Product and Engineering. Before I hand things off to you Gabe, I would just like to remind everyone that will be having a Q&A session at the end of today's talk. So, if you have a question throughout the presentation, please enter those into the questions tab of your GoToWebinar panel; you can see how to navigate to this on your screen here. All right, Gabe, I will go ahead and let you get started with today's discussion and thank you for joining us.

Well, thank you Delaina. So, I'm very excited to jump into this discussion today. We're going to talk about how to have an efficient and productive experience following the AMP guidelines using VSClinical. But before I do that, I want to acknowledge also that we have this work that we're discussing today, as well as a lot of other work at Golden Helix, has been supported by the NIH through the award of these grant funds. So, let me just read through this here for you. The research reported in this publication was supported by the National Institute of General Medical Sciences of the National Institute of Health under these award numbers and our PI is Dr. Andreas Scherer the CEO here at Golden Helix. The content is the sole responsibility of the authors and does not necessarily represent the official views of the National Institute of Health.

Alright, so let me give you just a little bit of a background as a company here. Golden Helix is a bioinformatics company started in 1998. We have been serving the research, and more recently the clinical markets, across all of these different context. And what we're going to be talking about today is that context of genomic testing and particularly the context of genomic testing and cancer. But we also support other types of workflows, whether it be research workflows around say diagnosis of rare diseases, other types of ongoing research workflows. But this work is generally going to be covered under the VarSeq product suite that covers a lot of the variant annotation filtering and interpretation starting from Next Generation Sequencing data. That NGS data can also be aggregated in our VSWarehouse where you can store the variants you've seen over time, as well as if you are doing sort of large-N analysis we have and a long history of supporting the research market with tools specific to things like microarray analysis and GWAS workflows in SNP Variation Suite, but we're going to be talking about VarSeq today.

So again, because we've been around for a long time and supporting the research and clinical market, we have a very vetted solution. We've been cited 1,000s of peer review publications and have over 400 customers globally. Now that covers industries such as pharmaceutical research hospitals, government institutions, but of course a lot of clinical labs as well. When you go with Golden Helix, you're getting very vetted software. Everything we do is immediately scrutinized by this wide set of use cases under this wide user base and we take the feedback of our customers extremely seriously. We're always looking to improve the software and go in a direction that matches sort of the way that people are pushing the boundaries. So a lot of what you're going to see today is again about the clinical labs that we have been working with, starting to do this work and we have been learning about how to make their lives more productive and efficient, and how to make it easier for a new lab that's getting started to bootstrap.

So, we are always in this mode of creating content such as this talk today, aggregating the best practices out there, and supporting the users on their work. And that support is really because we can be aligned with your goals as an institution, as a user, under the model of just having a subscription to our license not based on per-sample fee or something like that. So, you just have a subscription model with unlimited training and support and again, you get the vetted solution that you'll be seeing today. So, let's go through a little bit more on the overall workflow of NGS analysis in the context of a clinical testing lab. So, you're starting off with sequencing or short reads that are often selected to match only a set of genes for a given targeted gene panel. Those regions will be aligned, and those alignments will be placed in a BAM file, and then the variations to the reference sequence will be placed in a VCF file.

That's where we're going to start talking about today. If you need to have those steps also covered by your Golden Helix solution, we can do that with our partners with Sentieon. So, we are bundling and supporting a variant calling solution from Sentieon there. On top of those BAM files, we can also call copy number variations, that's going to be really critical and a cancer workflow. There's a lot of gene amplifications and duplications that are going to be important biomarkers for precision medicine. But I'm not going to have a chance to talk about the much of that today. We both sort of mentioned this but if you do want to get into some of these other aspects of our clinical stack, please check our archive of webinars or contact us, we'd be happy to give you a demonstration of this because it is a fascinating and interesting topic and we have a very unique and powerful solution there. Similarly, I'm not going to have a lot of chance to go into the aspects of all of the great annotations that you can bring in but we do have a full suite of public, curated, and licensed content that allows you to very quickly and efficiently get to list of prioritized variants that have the impact or the potential impact on a gene of interest - whether those are small variants like SNPs and indels or things like CNVs; you can do annotation and filtering workflows with VarSeq on both of those. Once you have a set of variants of interest, it's time to put those into a clinical reporting workflow. And this is where we can follow different industry standard guidelines that have standardized the kind of systematic process here. There's going to be a number of steps of that process, but the big one is essentially interpreting and scoring a variant. So, following the ACMG guidelines, we can score variants pathogenicity. Or essentially interpreting and scoring the clinical evidence that you might put in a report and that's what we'll be talking about today following the cancer guidelines based on AMP.

And once you have those highly vetted and interpreted variants, you can compose those into a clinical report and we have a fantastic templating system that allows you to customize that report to match exactly the content and boilerplate of your lab, while also being produced automatically based on what you work within the interface. You can produce that report in Word format, in PDF format and all of this information can also be easily exported so you can dump out Excel files that contain filtered and annotated variants etc. So that's at the kind of 10,000-foot view, again, what we're going to really focus on today is VSClinicals’ AMP guideline support. So let's talk about this product and we just came out with this product earlier this year, in the summer actually. And in fact, if you want to get a little more of a kind of comprehensive review of this, I suggest you look at our webinar archive where we cover the introduction of this and we also have a couple webinars since then covering things like somatic variants scoring, etc.

But the idea of this workflow is to follow the recommendations of this paper; Standards and Guidelines for the Interpretation and Reporting of Sequence Variants in Cancer. And this was a generally considered the AMP guidelines, but as you can see a lot of friends are involved here, so we have the ACMG group also supporting and authoring this paper. What we are able to do with this VSClinical workflow is really standardize your entire lab process of going through interpreting and reporting variants. You can build your customized reports. You can automate a lot of the steps that can be automated such as: variants scoring and matching the evidence that is out there all of these annotations that I mentioned, which one's match the current variant, the match the current tumor type, which things would you need to look at to follow these guidelines and have the guidelines sort of embedded in there. So, having things like the clinical evidence rating system, which we’re just about to discuss, built into the workflow. And finally what we're really going to focus on today is this product has a fantastic ability to reuse your interpretations and build your own lab specific knowledge base. And that's going to have a dramatic impact on your ability to handle higher volume of samples and have a reproducible result. So that's sort of what we're going to talk about in summary. So, let's get into this in more detail. So, the first thing is, what is this AMP classification system or AMP clinical evidence evaluation system? So, what's going on here is the guidelines are providing a structured rubric to evaluate the quality of clinical evidence that you might want to report. Otherwise, you might just find a piece of evidence and you put it in there and you know one lab says that this is meaningful and another lab says this isn't meaningful. But following the guidelines ensures that the information going into your report meets this industry-wide standard for quality and, this is sort of a key bit, the applicability to the patient's tumor type.

So, this Tier 1 evidence - this is about finding high quality evidence that directly applies or has been shown to be applicable to the current tumor type. While Tier 2 evidence might still be as just high quality, but it is for a different tumor type or this Level D here is preclinical information for the current tumor type but doesn't meet the same rigor in terms of the level of evidence.

So, next we have this information and we can classify the evidence that we might be evaluating. Let's go ahead and talk about how this goes into a clinical report. So, following the guidelines ensures that we have the right type of biomarkers that have this evidence threshold met, but how is this going to be presented? This goes beyond a bit the scope of what that AMP paper really covers, so we kind of have to go beyond that and do our own research if we wanted to discuss this. We want to report clearly what this actionable evidence is, but there's also information about drug sensitivity, drug resistance, diagnostic, prognostic evidence; this is kind of the information that we want to put in there. But how do we sort of describe this? You want to describe it in a way that is going to be readable by an oncologist who may not have a strong genomics background or other medical professionals that are reading this report. And so, they need to have some context, some genomic background, some references to go read, all that information needs to be available and written up in the report. Similarly, we're going to want to have things that are describing the variants at a technical level. But also, in a way referencing say, known genomic identifiers like dbSNP or you know, that's president popular genomic database like COSMIC and what those identifiers are if someone did want to follow up on the specific discovered variants. So, to get an idea of how we might structure all this information and what level of detail we want to go, let's look at some examples of reports from some successful genomic testing labs. First one up here is from the Memorial Sloan Kettering Cancer Center. These guys process tens of thousands of samples and they have this very popular target panel called MSK-Impact. In their report you can see that along with some variant descriptions and the relevant drugs that are here, there's also this descriptive paragraph. This kind of starts with a high-level description, may be describing the genes function pathways and the cell functions and then narrows down to affirming the biomarkers oncogenicity and the relevance of a class of drugs for this specific tumor type based on this biomarker. So that's one kind of nice brief and composed version, but you can already see that there's bits here that are very generic about the gene and there's bits that are specific about the biomarker and then there's bits that are specific about the sort of targeted therapy and the drugs that are listed. Alright, next one here is from Foundation Medicine. This is their flagship report called Foundation One and you can see they have this genomic findings section where they have all of the genomic alterations listed and for each genomic alteration, they start with these different, and they break this down by sections. So, they start with this background information about the gene and the about the genomic mutation and then they have this information here kind of covering the frequency of these types of mutations in this type of cancer, and maybe some information about the outcomes you expect based on mutations being present. And finally they get into the clinical evidence. And this is presented with the implications of what implications does have of having this biomarker in this tumor type on potential treatment strategies. So that gives us probably a higher bar then - if you wanted to sort of have a report that is really kind of written to have a broad audience you want to have probably this type of detail and this type of breaking it down to background information about the gene frequency and prognosis treatment strategies in the case of this biomarker.

So that's the goal, write up this interpretation about the AMP tiered clinical evidence while presented with some great and relevant background information that allows for you know, representing the best understanding of what is going on in the field of genomics and cancer.

But as you start doing this work as a lab, you'll probably quickly realize that writing these interpretations can consume more time than the rest of all the other hands on steps of the testing process combined. On the right here, and this is definitely a worst-case scenario, but we can see that we have over 1,000 publications that specifically cite BRAF V600E. Being the most commonly mutated recurrent cancer mutation, this is definitely the worst case scenario, but it gives you a sense of the scale of the work to write these interpretations and to sift through and understand, what is the best understanding of the field. Next thing is, you know, you're going to have multiple lab personnel working on this project of writing these interpretations. So along with the medical director, you might have multiple people that work on it during different times of the week, or maybe you have a switch off of one week to another.

In fact, I had one stakeholder that I was talking to an interview and they kind of, you know expressed this concern that you know in our lab if you send us a sample today, you'll get back a report. And if you send us that same sample, maybe two weeks from now, you might get back a substantially different report in terms of how things are sort of interpreted and described because a different medical director will be signing that out. And finally, you know, this type of work is very time consuming and it makes sense for some of it to be done by other lab Personnel than just the medical director. So, you might have a resident or some other technical technician working inside here and starting with sort of managing the data and checking the quality but also, you know writing some of these interpretation. So, then you know, you need some coordination of how this is going to be managed. How do you know if the interpretation needs to be reviewed? How does a resident sort of leave a comment for the medical director to take a look at something or how does the medical director keep track of what exactly in this report have I previously signed out in a previous report or what has been written up from scratch? So, there's this problem and thankfully, I think the solution can be quite clear.

Because in most tumors we're going to have genes and biomarkers that we could call frequent flyers. They're going to occur very frequently. They're going to have a very high recurrence rate. So, we just need some way to store and reuse our interpretations. So, you know, we see that biomarker in the future and that same tumor type, and all the work has already been done for us. We also be nice to support this coordination effort, so we know you know what things have been previously written and what things are new or have been written up and need to be reviewed.

So, the solution to it might be just then to save our interpretations into a knowledge base. But if we just do that if we just have the interpretation saved at the level of, I've seen this biomarker in this tumor type, you'll start to see a lot of you know copying and pasting because you'll have some information that's kind of at the gene level and you copy that next time you see a different biomarker in the same gene, or maybe there's some text that's really applicable to all tumor types and it's not specific to just say melanoma and then you'll have the same text being starting to being copied into the same interpretation. So, you have these parts that are shared and they really should be reused at a more granular level. So, to reduce this copying and to improve this reuse we break up our interpretations into these seven snippets at the biomarker interpretation level. So here they are here on the right.

So, the first three of these are purely descriptive and they're generally always going to be used and the other four are getting into that clinical evidence data if there is reportable clinical evidence. And in fact, with saving the interpretation here, you're also saving the AMP tier. So, tier 1 A, tier 1 B, etc. of the clinical evidence for given biomarker in the context of a specific cancer. Now we can also adjust the scope of these snippets that are saved and apply to future cases. So, for example, a description or clinical evidence statements may apply more broadly than just to a specific tumor type of the current patients. So, we want to be able to broaden the scope and say let's save this for all cancers or maybe instead of non-small cell lung cancer, let's save this for all lung cancers. And then also, you know, the specific mutation type may be too precise of a scope. Another example would be if you see a protein truncating mutation and you're saving an interpretation of what the impact of that loss of function mutation is for some tumor suppressor gene you want that interpretation to be reused the next time you see any loss of function mutation in a tumor suppressor gene, not just that specific one. So, you can change the mutation scope to be, let’s say all truncating mutations. Another part of writing reports is to provide good citations and we support this in VSClinical by every time we see a PubMed ID referenced in the interpretation text or even a clinical trial number we extract that; we actually go and look it up, we call out to PubMed ID get the Title & Abstract and we put it below just so you can see what you're looking at. And then we also add it to the reference section of your report. Finally, we want to have a system that supports this multiple user use case and it can grow and evolve over time. So we need a system that as you make changes you can see side-by-side differences highlight those that have, you know changed from your previous version, have some nice features to support this collaboration, things like that and we're going to talk about that and it's better even to show that and we'll show that in the upcoming demo.

Most importantly, we don't want you to have to start from scratch. So, your own lab, you don't have to start from scratch because we have written a lot of these snippets for the most common mutations, these frequent flyer mutations in these frequent flyer genes for the most common tumor types. That means that when you add your first biomarker using VSClinical you're probably going to be at a really good starting point. So, let's talk about that next. It's important to emphasize what this can actually mean in terms of your overall productivity. So when you go into VSClinical you can not only get you know, your annotated and clinical evidence pulled in, organized and help you write your interpretations, but you have the productivity enhancement of our ever-growing interpretation knowledge base and we're calling that Golden Helix CancerKB. And then you also have these collaboration tools in this infrastructure so that you can use that as a starting point and build your own lab specific knowledge base.

That means you know for both samples that contain biomarkers you previously seen, your time can from spending hours writing up these interpretations to just minutes, to just review what's already there make sure there's nothing sort of unique or that needs to be updated our CancerKB not only includes the snippets, but for those clinical evidence sections as we mentioned it's going to have these AMP tier classifications and a list of the relevant targeted molecular therapy. So, it's going to have those list of drugs that is referenced inside the interpretations. So this information is useful, not only if you have that exact biomarker that we interpreted but we actually have on the right hand side this display that will show you other interpretations coming from CancerKB or even your own knowledge base that are for a different mutation in the gene that you're looking at or maybe it's the same mutation, but it's in a different tumor type. So essentially sort of these nearby things and that can help you get started when you're doing something new but it's still somewhat like something that's already been done.

So, as I mentioned this is going to be an evolving thing currently CancerKB has over 200 insert, 60 of these interpretations snippets and over 64 gene tumor pairs and 30 of these clinical evidence statements that have reached a tier one level. So, this is describing this kind of FDA approved drugs for these well-established high-quality biomarkers. When you see new or updated interpretations and you're saving those to your internal knowledge base, there is an option to share that information back to our Golden Helix team and then our curators will review that and integrate that with everything else that they've received as well as new content coming from the world in the next version of CancerKB. So, this gives CancerKB a really great chance to kind of grow as we do the work, to put in, to keep it up to date, as well as what you share back to us.

All right, I think that's a fantastic background of what we are talking about today. But overall, I think it's always best to see this in action. So, I want to take you through a little bit of a workflow here in VarSeq. I'm going to switch up here. And what I'm going to do is I'm going to act essentially as the medical director here. I'm going to open up a project that's been already been started by a resident.

So, opening up this project I can kind of get a sense of where this was left off. So, in this case, I'm starting with the first sample here and this sample has a number of variants that have gone through a very straightforward filtering process. Again, you can customize this and run the exact same steps every time you see a new sample in your lab based on whatever your level of detection is, whatever your filtering strategy is. And this case I'm just looking for variants that are of reasonable read-depth, have a reasonable level of presence at the variant level over 1% here for example, have some impact on the gene a loss-of-function mis sets or splice like prediction and are not in Nomad at 1% or more and that are present in COSMIC so have some sort of recurrence as a somatic variant. I can look at this in a tabular form and see all those annotations.

I can also look at the coverage of the targets that I have my amplicon base panel in this case based on. Now thankfully, I don't have to go and look at this table, although you could. You could set up filters and look at this table all you want but I'll show you in a little bit that we actually summarize all this information inside our VSClinical workflow. One thing that my resident has already done is gone through and selected a couple variants and kind of for each variant did a little quality assurance and there's really no real substitute for the amount of fidelity you have by using a visualization tool like GenomeBrowse to show you things like, here's the called variant, here is the coverage, here's my target, here's the nearby variants in COSMIC. You can also show things like you're nearby variants in your own catalog or your warehouse, etc. and I can kind of inspect this very straightforward SNP here, but I can see where the reads are and what depth etc. and even pull up a little chart. For things like insertions and deletions this is kind of critical also to ensure that you're not looking at false positives that the variant is being well understood and that case I've already gone through again. My resident has already gone through and done some of this work. So I'm going to switch over to the VSClinical interface and I'm going to hide my “Filter” interface for now. And I'm going to work backwards. Let's go to just see how much work has already been done for the sample by going to the “Report” tab. Now, the “Report” tab, and I'll go through these tabs a little, bit but the “Report” tab essentially the combination and the outlining of all the work that has been done so I can see the sample and patient information that was put in the “Patient” tab. I can see that there's already a biomarker that's been added. We have V600E here and some interpretation there as well as lots of other details about a secondary finding, etc. So I can look at it here. I can even click on these things and review them as well here. But in this case, I'm actually just wanting to say well, okay. Well, what's this look like in the draft report so I can go to this exports list here. I can see this report in PDF view form could also open it up in Word, click on this word icon to open that - yep, and I can see that we have, you know are bereft v600e biomarker. And this is that type of interpretation level very similar to that Precision medicine report where we have clinical outcomes and frequencies drug sensitivities clinical in for sort of technical information describing the variant. And in this case, we also have a variant coming in from our ACMG guidelines, a variant of unknown significance. A summary of our NGS coverage, a list of regions that fell below our target thresholds, and of course these citations and descriptions as well as the inline references that were part of those interpretations that we're seen above. Again, all this report information is generated automatically as I go through and update anything. I just want to I just click this re-render button renders that report I can click it and open it and see that updated information.

So let's go through a little bit of these tabs to see what has already been done for this patient before I sign it out. So I have my information from the sample and patient information. I also have a tumor type or current diagnostic set. And this is a melanoma sample. Why is this important? Our AMP guidelines specifically state that when we interpret the clinical evidence that has to be in the context of a given tumor type. So some evidence that might say that this is a Tier 1 drug for melanoma would actually be a Tier 2 level drug for non-small cell lung cancer. So that's why we need to be upfront setting this information and being very clear about that. All that information that I saw in those other tables can be summarized here about how many variants are present in the sample. How many have the low allele frequency? Our coverage Target regions? We have 500 of those how many are passing or thresholds here. Here's a few that are not passing. And I can dive into those as well as look at the summary information about each gene that's covered. Real quickly, I'm to describe this must call hotspot section.

Now you might as a lab that it makes a sense to never sign out a report for a melanoma sample unless you know you have good coverage over the 600 position o you're not getting some false negatives there. You're not missing the presence of a mutation just because say a target failed and you didn't get good coverage. So in this case we can see that we have fantastic coverage over this position, and I can actually see that over here, and I can even click on these to get sort of, you know exon level reports of these regions for these genes. Let's see what we have in our mutation profile. So this is kind of what I expected given that I already scrolled through the report, but we have one kind of secondary germline finding discovered. So while we were sequencing the sample, we found a variant in a heterozygous state that might be an important to inform the family of potentially hereditary disease risk of getting maybe breast cancer, etc. And that's already been established as pathogenic following the ACMG guidelines. Again, I don't have time to go into following those guidelines, but that whole guideline workflow is supported by just going to this “Variant” tab and following the guideline workflow specific to this origin type. So in this case a germline variants would follow the ACMG guidelines and that's going into our secondary germline section of our report, versus our biomarker section, which we would describe for things that have direct implications to the therapy or outcome of the patient in the cancer context. We have our V600 variant here and we can see that that's present at about 6% of the reads and this again follows a scoring system. Again, it's applied automatically a lot of work is done to really support deep-diving into variants especially ones that you haven't seen before that maybe aren't the frequent flyers. This being a frequent flyer variant, it really does hit all the marks of being an oncogenic variant. Again, we have another webcast where we go into describing how we do this variant scoring and classification, but now that we have decided that this is oncogenic we're report it as a biomarker. And today we're going to spend all of our time in the “Biomarker” tab going through that evidence there.

All right. Next thing we're looking at here is the MPL variant which is a missense variant, but it does not meet our threshold for being oncogenic. So it's going to go into our variants of uncertain significant section of our report. Similarly. You might evaluate a variant and be like, you know what it doesn't even make sense to do that. It's a benign variant in the could just dump it into “don't report”.

So let's see what are our work has been done so far on this BRAF-V600E we can see that overall. It has a tier one level a in one of its clinical evidence sections. Although it does have two different sections that have been filled in drug sensitivity and prognostics. Now, it looks like there's unsaved changes in drug sensitivity. That's what this summarized over here is one unsaved and our overall interpretation sections gene summary etc. looks pretty good. This biomarker summary looks like there was flagged here by my resident wanting me to take a look at that. So we can go ahead and switch on over to the biomarkers tab. I can click on this even just to jump straight to that could see my tab going over here, and this tab is oriented around providing one section for each of these different interpretations snippets and each one of these has ability to be saved to be changed to be noted to be commented on independently. Now a lot of the times once you set a gene summary, you're good to go you don't have to think about Gene summary of anymore, but every once in a while this will updated and changed over time. So since we're going through this for the first time, I'm going to go through this again one by one - at least going through a couple of these. But just remember that overall you might skip most of this because you've already done the work that's already been filled inm you don't have to review itm and you just focus on the clinical evidence section i.e. what sort of the impact that I'm going to report of this biomarker in a clinical context. So since we're in BRAF we give you a bunch of link outs to be BRAF you can also go search all these fantastic databases. If you want to go read about BRAF and these cancer specific databases. We also provide again value of VSClinical not only do we have a place for you to write this interpretation, but we give you this great deep dive into all of these different genomic annotation sources and what they have to say about BRAF - so COSMIC has a nice functional summary. They also go into sort of the role of cancer and specific Hallmarks of cancer and how BRAF sort of impacts those and Times calling out specific tumor types as well. Similarly. We have high level descriptions coming from CiViC genetics home reference. This is a little more high level background information. But I do find that sometimes depending on who you want the target audience for this to be a fantastic resource where you might see these great descriptions of how often mutations occur, etc in a general sense. CGD, this clinical genomic database also has high level descriptions that are very useful as well as NCBI. So let's take a look at this section. Now notice I have the scope here set for melanoma. What that means is that there might be a might want to for example also in general you might set BRAF description for all Cancers. And so the next time you see a BRAF in any cancer, that would be the scope. But in this case, I wanted to write something specific to melanoma. It turns out CancerKB also has an interpretation over here. This is that sort of relevant interpretations for other tumor types. This is a BRAF description for all caancers that has been updated and you can see also the exact details and references from that description. Let's look at our history. So it looks like this started off being filled in by CancerKB and then my resident Allison added in a comment that she added into something about a knockout reference. Okay. So it looks like there are some extra words and that was saved recently to our history. So you can see our version history has two entries. Let's take a look at what those entries show. Okay, right now everything matches exactly to our saved interpretation. You can actually go ahead and start doing some edits here and you'll start to see, you know, if there are any changes here that you'll see like I'm adding a new line of text or I'm deleting a line of text Etc. But also you can see the differences from the original entry, the one that was filled in from CancerKB and it looks like what Alison did was add this extra line. Maybe I don't want that line. I can go ahead and delete it. And again, we'll see the differences against that will show up as a nice big red blob Etc. so just to something to keep in mind. It's easy to sort of keep these histories going if I want to I can say review and save now I'm going to save my change here again. I have that option to share my interpretation back with the Golden Helix curation team going to go ahead and save that and now I have my history which says I saved this and I can keep that around and keep track of that and that's what we're going to use for the report. So that's saved now that's completed. Now. There's no longer sort of an edit signal or anything here. Now, let's go through this next section - alterations and frequencies. The idea here is we want to have a kind of a context about how often are mutations in this gene impacting this this, you know, this tissue type this tumor type. And what is the sort of general outcomes for having mutations in here? In this case we have a description that's already been filled in without being edited directly from CancerKB for melanoma. You can see there's also a different description here that's been written up for non-small cell lung cancer because there's also a different set of potential implications and outcomes for non-small cell lung cancer. We also have from COSMIC and MSK a little summary of all the mutations from cosmic and which which tissue types those mutations are what, you know, the tumor types of the mutations that are present there. So you can see that there's quite a lot of skin and a thyroid cancer that have BRAF mutations. Alright, so let's go ahead and move on down take a look at the next section here, which is are BRAFV600E biomarker summary. And this had a section with a little flag. Looks like Allison was wondering if we should add some sort of paper reference. Notice, there's no inline citations here and I agree. We can look at our nearby interpretations. It looks like we have interpretations for V60 K and other non v600e mutations coming in from CancerKB, that's interesting. None of those have interpretations or papers per say, there's also a description of v600e and non-small cell lung cancer, but that's a paper that's not relevant to what I'm looking at now. Let's look at some of the information we have about this variant coming in from VSClinical. This is that Oncogenicity scoring system - summarizes a lot of the points about why this is an oncogenic variant. We can look at linking out to ClinVar or Cosmic if I wanted to, it's as simple as clicking out and going to those individual pages. So I could go look at COSMIC for this variant. Could also look at my insilico predictions. I could go look at the 1,000 publications here, not going to do that right now. I could go search Google Scholar. For example, go find v600e in public papers. That might be something I go Do. We also have this “Assessments” tab and this just brings in specific detailed assessments about this mutation from different sources. For example, we can see PMKB pulls in here a description with some useful papers. And I think this is a good paper. Let's go ahead and add that as a reference here and I'm going to just go ahead and have that. I'm just going to hit paste and as I put in that reference, you can see just by having that PMKB reference here. It pulls it up pulls it as an inline reference will be added to the report if I click on this I can actually see the abstract and title and everything. So it gives me a great context for being confident. I want to add that reference. And of course I might want to add another sentence here, but at the moment I'm going to go ahead and remove that flag. So I'm happy with this and so I’ve unflagged it. You can see the kind of history of my collaborative progress on this section.

Now finally, this is sort of where the meat of the work is done. All that rest of that information is really just teeing up the critical information to describe What is the clinical evidence for drug sensitivity, drug resistance, prognostic, or diagnostic? So each of these tabs is filtering data coming in from different sources. So this is coming in from Civic vs. this information is coming in from maybe Pub PMKB and the descriptions that they have here, prognostic evidence drug resistance evidence coming in from Civic and we have star ratings, and Drug sensitivity evidence. Some of this is coming in from DrugBank. So we have very good high quality FDA approved drugs of various types and combinations to see here as well.

I'm not going to go through the interpretation and writing process as we're really just focusing on saving this information and seeing how we can work in a team to do that. So in here I can see that I've already gotten a couple drug combination therapies selected. We have our Tier 1 level A FDA approved therapies, including a professional guidelines for this tumor type and we can see what are our interpretation is here. Looks like my resident had made some changes. Let's take a look at what those changes are. In this case she referenced the fact that there's an ongoing clinical trial that's very relevant that I would like to you know, put in the report and notice by just adding this clinical trial this NCT ID here - that actually also goes into our inline references so we can reference PubMed IDs. We can reference clinical trials. I can say let's go ahead and view these and I can look at that clinical trial information and it’s even got the embedded inclusion criteria in summary coming in from clinicaltrials.gov. So fantastic information, this all looks good. I'm happy with this I can go ahead and say let's go ahead and review and save this. So now I have that covered. And finally I wanted to review and save my biomarker summary. So I have that covered. There's basically no more to do’s is left on this now actually before I do that. I just want to go back to my “Report” tab so you can see that context over here. When I'm about to sign out my report, I have one to-do left which is that one last unsaved interpretation. So we don't let you sign out until you either choose to save things or maybe not save things. If you don't want to save them that's fine. We have this option here that says don't save - maybe you're making a change that's just specific to the sample, but just be aware of the you know, if we want you to be aware of are you writing interpretations that you want to save for your next time do that now and then you can sign out your report. So, finally I can look at my final summary going to put in that. I I'm fine with a summary here. We're going to go here and sign out this report and this sort of puts everything in freeze mode once I sign this I can no longer make changes to any of my interpretations. It puts this in a read only mode, but it does capture all the work that was done and I have this project I can come back to and see exactly what happened here in the future. So I'm going to go ahead and confirm that now it says that I sign this off has the report date. Let's go ahead and update the report here. I'm going to click render and open that up as a PDF and sure enough It's no longer in draft. I have when it was signed off. I have who signed it off and when they signed it off Etc. So all that information including those all those updated references and clinical trials are in my report. Can save that out as a PDF Etc.

So let's go ahead and switch back to the slides and we'll go and just cover a little bit of background or wrap this up with acknowledgement as well. I want to always keep track of the fact that we are grant funded on this work. And so I want to in summary just say thank you to the NIH for that grant funding and we also want to take your questions and have time for questions. I realized I didn't have a lot of time left and I wanted to get back to that. But one thing I want to do is just give Delaina another chance to give you an up-to-date status and where things are at in the summer and the fact that we just released this product VSClinical AMP workflows and give you an opportunity to hear about that as well. Great. Thank you Gabe. And we will go ahead and give everyone a second to enter the rest of their questions in that panel and as we wait I have just a couple of housekeeping items. So first up is our Summer AMP Sale! If you haven't heard about this already as the name suggests This is our Big Summer AMP Sale all focused around the AMP workflow. On the screen you can see the different bundles our team has put together and these are at heavily discounted prices and then on top of that as Gabe mentioned earlier in the presentation, we are on an annual license business model. So the thing to note here is that all of these bundles are 15 month licenses. So on top of those low prices, you also get an additional three months.

And then since this is our big summer sale our team said why stop there and they added a third year for free if you purchase two years! So lots of stuff happening there, but what there is not a lot of is these bundles so if you look on the bottom of each of these bundles, you can see the number remaining and as you see there, there are only a few left - one even has zero remaining. So if you're interested in any of these, please reach out to anyone at the Golden Helix team and we will go ahead and put your name on one. This will be ending on September 30th. So definitely make sure to do that before then.

And then secondly, we will be heading to ASHG 2019 in Houston, Texas in October. So, coming up relatively quickly, mentioning this now so you can hopefully add a stop to the Golden Helix Booth into your schedule. Our team would love to connect with anyone who is attending so definitely let us know if you're going. We have a lot going on at the show. We'll be doing our in-booth demos. We have several CoLab presentations on the exhibit floor each day. And of course handing out those infamous t-shirts that you might have heard about or seen on Twitter. So please stop by and say hello, or if you would like to schedule a one-on-one meeting with our team at the show you can email info@goldenhelix.com. Perfect. All right we can move into questions:

Question One: Does the VSClinical workflow support gene fusions?

Answer: Yep, and if I had more time I could get into another example of that but not only will we have gene fusions but you can add your own gene fusions. We have some of those in CancerKB as well, especially if you can think about some of the ones that have some pretty strong clinical evidence ABL1 type Gene fusions as well for example, and similarly your CNVs. As you add CNVs into your workflow, some of the more common CNVs are going to have known clinical statements as well as biomarker descriptions and some of those have emptier classifications.

Question Two: What about common drug resistance mutations, like EGFR T790M.

Answer: Yeah, and that's where it starts to get into again. You want to have a tool like this that can pop up those types of maybe less commonly seen but very critical to understand variants. So not only will our oncogenicity scoring system score those very high based on the cancer evidence available, but we actually have interpretations in CancerKB for a lot of those drug resistance and we separate out drug resistance versus drug sensitivity and to two different clinical evidence sections. Sometimes it was sort of merged together and they can be kind of blurring the lines there. But we consider that a distinctive enough difference to have a different section describing like a drug that you might be sensitive to or that has some sort of resistance built up because there is a T7ADM mutation that has grown and in prevalence at a higher allele frequency. So that's kind of an important distinction. Question Three: Can this workflow be automated so that we can do it in a larger higher capacity setting?

Answer: Yes. A lot of what you're seeing here, we obviously we can't automate the hands-on interactive steps, but getting to the point where you have that project that's fully constructed where you have all your variants annotated and filtered and selected to match your labs filtering process and have all the BAM files have all the computations done on those your cnvs called your coverage called all that can be automated in the command line process. We actually had a really popular webinar earlier this year talking about automation. So I would suggest you go look at that. If you want to kind of get a sense of how to take it to the next level in terms of having a fully automated lab process.