 Okay. I was tasked to react to Gail today, so I don't know if it's a react or respond, but the way I look at it is she's coming from a clinical geneticist point of view. I'm coming from a clinical laboratory point of view, and I don't have any disagreements with her, but maybe I can kind of put things in my perspective, and the examples I'm using today are the examples that I happen to be dealing with when I was putting this presentation together. It's a snapshot of maybe a one or two-day time period in my life. First of all, those bake-off results, this has been so helpful and so useful for us. It's very much encouraging, but yet obviously improvements are made, and they're talking about the differences. I use one degree of separation or more than one degree of separation to try to describe these. We now are starting to see what the differences are, and one point that I would say is maybe the Cicero and others could then talk about what is our target performance, and how many do we want to have less than or equal to one degree of separation. That's the goal and are we ever going to get there, and how do we start measuring these types of improvements? Gail also mentioned really the need for the follow-up and identify any problematic or misunderstood or I should say misapplied of the evidences from the ACM GAMP guidelines. And what I have found when there has been a major discrepancy, when I get on the phone with the other laboratory, they often have internal information that I don't have that isn't shared, isn't publicly available, and once we share this information together, we can often come to an agreement. Here is just something from Clint Vare of an uncertain and likely a benign and a pathogenic. This was fairly old, and I will tell you this is actually ARUP. We submitted that a number of years ago, and we need to be updating it. However, it's difficult to keep the resources to continually be updating this as well, so it's not only adding information to, but a continual updating of new information. One of the major problems that I'm dealing with today is that we get requests to perform targeted mutation or targeted familial variant test on an individual whose relative was tested in another laboratory. And we're getting reports that this is a pathogenic variant, and now I'm supposed to test for this so-called pathogenic variant in this individual, and when we look at the evidences, we don't agree with that classification. So how do I write a report, and those of you in the room that have tried doing this, you will sympathize. How do I write a report, whether or not the patient, this individual I'm testing is positive or negative, to basically say, we don't really think that this is pathogenic, or we don't know yet if it's pathogenic. Often the problem is that the clinician has misinterpreted a likely pathogenic report to be pathogenic. Other times we have the report in front of us, and a statement. Our laboratory follows the ACMG guidelines for variant classification, and I looked at this and I said they may be following them, but they're interpreting them very, very differently, because if I'm following them as well, I would never have reached pathogenic with this from the way that I interpret them. So not only do we need consistency in variant classification, we need to start showing transparency in how we got there. It didn't help me to hear this lab-followed ACMG-AMP guidelines without knowing exactly what they used. So as has been mentioned several times this morning, low penetrant, mild variants, or the combinations of these in recessive diseases are still difficult, and the terminology even itself we could potentially get better at. So for example, and I use CF and pancreatitis, it's a combination of mutations. Some would be mild, sometimes they're called mild mutations, sometimes they're called low penetrance mutations, sometimes they're called a riskalil for pancreatitis, and yet there's also different definitions of what a mild disease versus an atypical disease is. So for cystic fibrosis, a mild disease is still cystic fibrosis, but without pancreatic involvement. Yet an atypical disease would then possibly be pancreatitis. So we could use a lot of help in just defining some of these terms that we can all start talking about the same things. Now we really haven't dealt with riskalils, and I have to admit I don't do too many riskalils from GWAS studies in my lab, but we do do a few, and they are going to be more tested. So we do need to start addressing those. I know that there are efforts underway for somatic variants, for in a sense take the ACMG and P guidelines into the somatic variants, and it's very much needed, but it also shows you the real differences between the somatic and germline interpretations. So one of the areas that I really would focus on and push for is annotation justification. How are we justifying this? Here is a twin var entry for a variant, and I chose one that basically there wasn't any disagreement, but out of the ones that submitted, they were all clinical laboratories, they all agreed, and they all have put in their criteria for how they evaluate, but only one of them really gave a description and told me which publications they used to do the classification. So yes, I'm going to embarrass you Heidi, there's no way to cover this up, but I can't say how much I appreciate this because then I can compare the publications I found with publications that this laboratory has found and make sure that we are in agreement and at least looking at the same data. And also what has been brought up is a case-level data, and this is something that I just dealt with last week, an MSH6 that was previously described as pathogenic. We had then received a relative to test for this known pathogenic mutation in a relative, and we decided it's been a while since we looked at this, we revisited and sure enough it's been downgraded from pathogenic to a VUS in Insight. Very much appreciate Mark Greenblatt because boy, he answers back quickly when I send him an email and ask him about it. There's been three reported patients with suspected constitutional or biolo-like mismatch repair deficiency. This is the information that we had several years ago, but as you look at it, they are not well described. We had one patient that fit exactly the same description in terms of this variant and another frame shift mutation with cancer in teens, early 20s. So it fit it, but on the other hand, it hadn't been described alone. And one important point is that if you followed the families and the lineages where this mutation or variant was coming from, there was no family history of cancer. So all of a sudden, not all of a sudden, after looking at this more, we realized we really don't know what it means by itself is this mild. It does look interesting that it seems to be appearing when there's another clearly pathogenic mutation in younger onsets. We do have one case that is consistent with it. So again, instead of three, we now can give you four. And we can start counting on this, but it's also going to be important to start looking at these mutations by themselves and not in combination. And when we contacted the author of the publication that had first then called it a VUS, her comment was, it's an enigma. So I guess VUS is the appropriate one. So the point is with that is that there's no substitution for collecting case and case and case until we can get a clear understanding of it. Now, we are often asked about reanalysis of the data or reinterpretation and I'm going to put them as two separate. One is reanalysis, take the data, you have new clinical information you review it again with your variant file and see what else could be there. But at what point do I say we're not stable yet. This is not a mature technology to the point that I can use it five years from now. I may or may not be able to. So one of the things that we, that I could use guidance on is at what point do we then say rather than reanalysis, we need to retest the patient. Now reinterpretation is different. So that means we now look at more information, see what other evidence they are and then am I going to change classification. We've struggled with this and what is our duty to re-contact patients. We attempt to but I will tell you as a reference laboratory it is very difficult to get back to the ordering physician after three to five years especially. So we need ways to do this without putting more burden on clinical laboratories to say this is your responsibility. So I'm looking forward to how CCR and others can help us solve this. Now one of the main topics in a hot spot right now is really the clinical validity or the utility. I think Gail mentioned gene lists and yes I mean even an exome now or a medical exome which is a very large gene panel. But maybe but those are being done and so the question is we have an exome, we have a medical exome, which do we do? How do we do this? We need data, we need these laboratories or the people doing this to pull everything together and so we can make decisions based on actual data. The other point I'm very much looking forward to the clinical clinician bake off in terms of the understanding of what we're getting. I need to find some time today to make a very difficult call to a clinician who called and was very frustrated about a genetic report that she received. It wasn't through our laboratory but it was through a send out of ours and so she's asking me now to interpret the report of another laboratory and there is one mutation and two variants of a non-significance in a gene that could possibly explain the patient's symptoms. My point is she obviously wasn't a geneticist because the comment that I have was she wants to know well is my patient pathogenic or not? But okay that is a little too simple of a question and those that are the specialists in genetics know that so how now do I explain this to her and help her along the path to get the appropriate follow up for the patients. I'm also interested in a clinician's understanding of the prior probability and exactly to make sure we're not doing circular arguments of this is it we found a mutation in somebody who has some symptoms versus we found a mutation in somebody who has a clear diagnosis with other bits of evidences and the last thing is whether we are now going from affected populations into healthy populations so newborns newborn screening, carrier testing for parents but also just a general population in an interest of saying I want to know what is in my DNA not necessarily for me but for offspring how do we do it right to make sure we're not causing undue alarm and getting more and more medical procedures on a patient than it really isn't necessary but also if we do identify it how do we make sure that they then become get into the medical system appropriately as well. Now I had a chance to talk with this doctor Keiku who is a physicist and he gave a very nice presentation of where we're going to be in medicine in about in the year 2050 and he talked about all of these different things but he didn't talk about genetic testing so I asked him about it and I said what about the interpretation of the genome and it's not a direct quote but the general comment was nothing will replace careful scientific valuation one gene one variant at a time and that's where we are and let's get going on it.