Welcome to the ACMG Genomics Case Conference. Today's topic, New Disease Gene Discoveries, Genetics, and Functional Evidence. This webinar today is hosted by Baylor College of Medicine. And this activity is supported by an unrestricted educational grant from QIAGEN Bioinformatics and the Ingenuity Clinical Decision Support Platform. As a disclaimer, this activity is only meant and strictly intended only for educational purposes. Presenters today are Xia Wang, who's a visiting assistant professor in the Department of Molecular and Human Genetics at Baylor College of Medicine, Dr. Chunan Chen, who's a clinical postdoctoral fellow in the Department of Molecular and Human Genetics at Baylor College of Medicine, and myself, Christian Schaaf, I'm a professor at the University of Cologne Institute of Human Genetics. Our disclosures for today's activity, Dr. Wang is a stockholder and has ownership interest in AI live diagnostics. Dr. Chen has no disclosures. And myself, I receive royalties from the Springer Publishing Company. There are also disclosures for you to read from the members of the planning committee and the staff of the ACMG. You can receive educational credit for today's activity. The American College of Medical Genetics designates this live activity for maximum of one AMA PRA Category 1 credit. In order to obtain credit, if you purchase CME at the time of registration, you will receive an email within the next 24 hours of the live event that provides you with a link to complete the CME certificate. Once you complete, you will receive a second email containing the certificate. In addition to that, everyone is asked to complete an evaluation. You may click on the evaluation link at the end of the webinar or access the evaluation via a link that will be emailed to you. These are the learning objectives for the overall genomics case conferences. And for today's activity, we have specific learning objectives. First, we hope that you'll be able to prioritize genetic and functional evidence to better understand disease-gene relationships, that you'll be able to name at least one database that provides evidence for disease-gene relationships, that you'll be able to elaborate on the ABL1 gene and its role in two entirely different genetic disorders, and that you would feel comfortable in discussing how in vitro assays can be used to provide additional information on variant pathogenicity using ABL1 and TRAF7 as two examples. If you want to ask a panelist a question, we ask you to please type your question into the Q&A box in the lower left corner of your control panel. The questions will then be answered verbally by our panelists at the end of the presentation. So, let's start the webinar, and let's start with two polling questions. I ask you to participate in those. Let me start those. The first question, have you ever made a novel disease-gene discovery or contributed to one? Option one, yes. Option two, no. Fantastic. And we see that basically half of you have already contributed to novel disease gene discoveries and half of you haven't. So a remarkable number of participants who have already made novel disease gene discoveries. We have one more polling question. What do you consider the most critical step in discovering a novel disease gene? Option one, finding the right patient. Option two, a thorough clinical phenotypic description. Option three, access to a large patient database. Option four, in vitro functional assays to prove variant pathogenicity. Or option five, animal models to show relevance across species. Obviously there's not only one correct answer for this, but we want to see what you consider the most critical step in discovering a novel disease gene. Please type in your answers now. Excellent, thank you. And it's interesting to see that obviously people have different opinions, but the majority of people find it critical that they have access to a large patient database and also in vitro functional assays and a thorough clinical description come in as second place in the importance for novel disease gene discoveries. Okay, so let's continue with the webinar. This is a slide that was provided to us by OMIM a couple of months ago, and it shows how far we've come as a genetics community over the last 15, 20 years, a remarkable number of novel disease genes that have been discovered. But you also see that, you know, this graph is not even plateauing yet. There are obviously many, many more disease genes to be discovered when we think about the human genome with 23,000 genes. We only have about 6,000 disease gene discoveries made and listed in OMIM right now, so much more to be discovered, and you can be part of that. But you know, how do you get there? How do you move from a single-patient screening clinic to a new disease gene discovery? In our webinar today, we will outline two examples of how disease gene discoveries have been made by our groups, and we will move you through from the clinical description and the initial patient screening clinic to the actual paper that has been published. So let's start with our first case. This was a patient who presented to the genetics clinic, a five-year-old male with a reason for referral being failure to thrive. He also had a large VSD, ventricular septal defect, mild coarctation of the aorta, and skeletal abnormalities, including scoliosis, significant pectus excavatum, which is a chest deformity, fifth-finger clonodactyly on both sides. He had minor dysmorphic facial features, and a history of congenital diaphragmatic hernia, and a history of chronic constipation. He was the only child to his parents and the only person in the family to be affected with this. He underwent trio-holexome sequencing, and a candidate disease gene was identified. This patient carried a de novo missense variant in the ABL1 gene. Now, many of you may be familiar with the ABL1 gene, but this really came as a surprise and, to be honest, not a very strong candidate to be considered to explain the phenotype of this patient, because all that we knew at the time was ABL1 is a very well-known classic gene in human genetics, known as a proto-oncogene, and most popular for being part of the fusion gene, BCR-ABL, as part of the Philadelphia chromosome of chronic myeloid leukemia cancer cells. Inherited germline mutations or missense variants in ABL1 had not been associated with any genetic disorders at that time. Our patient carried a de novo variant in this gene, and it was the only coding non-synonymous variant that was de novo in this patient. So this is where the larger genomic databases of more individuals, ideally thousands of individuals, comes into play. And we found in our clinical database of clinical exome sequencing cases that the same exact variant was also present in three additional unsolved whole exome sequencing cases. One of those three was also de novo, and the other two were inherited from a parent, but going back in the clinical history of those families, we found that those parents who carried the variant in the ABL1 gene were also reported to be symptomatic for the same phenotype that was of interest in the patient. In addition, there was one additional unsolved whole exome sequencing case with another de novo missense variant in the ABL1 gene. This really shows that, you know, one case is not enough, and sometimes you do need the power of large databases and large cohorts. If you do not have access to those kind of cohorts at your institution or through collaborators, there are also alternatives that many of you are very familiar with. One of the ones that we use most frequently is GeneMatcher that allows us to be matched with other researchers around the world who have discovered variants in genes of interest that may not be known to be diseased genes at that time. Having identified those five patients with variants in the ABL1 gene that were either de novo or segregated with the phenotype within the family, we took a genotype-to-phenotype approach, going back in the clinical histories of those patients, looking at their phenotypes. And we found out that basically all of those individuals had congenital heart disease, ASD, VSD, or aortic root phenotypes. Almost all of them had skeletal abnormalities, including pectus excavatum, scoliosis, finger deformities such as arachnodactyly, clinodactyly, or contractures, and also hindfoot deformities of the feet. Many of the individuals had a history of failure to thrive during infancy and early childhood. Given those phenotypes, the differential diagnosis that was listed for many of those patients included Marfan syndrome, Loewy-Steed syndrome, Schwentz and Goldberg syndrome, which makes sense to all of us as we look at the phenotypic descriptions of these patients with congenital heart defects, abnormalities of the aortic root, skeletal deformities such as arachnodactyly or contractures. So we actually went back to make sure that none of those patients, none of those five patients had variants in any of the genes that would explain or be causative for those syndromes. And we looked for the genes that cause Loewy-Steed syndrome, Marfan syndrome, Bencki's disease, Schwentz and Goldberg syndrome, and others. And those areas that were not covered by whole exome sequencing, we did Sanger fill-in to make sure that we really and truly covered 100% of the coding region of those genes. And none of the individuals that we are looking at had any pathogenic variants in any of those genes. So let's go ahead and look at the pedigrees of those families. There are four families that I just mentioned and one family that is unpublished, for which I will only present limited information. And you see we have highlighted mutation one, the more common missense change as M1, which is seen in red in those pedigrees. And you can see that for those individuals who had the phenotype within the family for whom we had access to their DNA, the variant segregated with the phenotype. And I will show you some clinical pictures of those patients. And you can see that they do share some features, some dysmorphic facial features. It seems to me that the young children, the infants, have a more round face with chubby cheeks, a pointed chin, and a very prominent forehead. And as individuals get older, the face kind of elongates. They have a long face as they become adults with a long and narrow kind of pinched nose, which seems to be one of the common dysmorphic facial features in the adult patients. In the next slide, you can also see some of the skeletal malformations that we see in those individuals, including the chest deformities, the scoliosis seen in the patient in the middle here. And you can see the contractures of some of the hands and the hind foot deformity with the medial displacement of the medial malleolus in the patient on the left. So to summarize our clinical presentations in seven individuals who have missense variants in the ABEL1 gene, we see dysmorphic facial features, skeletal abnormalities in all of the individuals. We have congenital heart disease in six of the seven individuals, joint laxity, contractures, or costochondritis in many of the individuals, a history of failure to thrive during infancy in six out of seven individuals, and frequently gastrointestinal problems. But those vary in presentation, and they can include constipation, reflux. One person had a history of intestinal malrotation, and we had individuals with colonic dysmortility. What I did not show and did not tell you about so far is that four out of the five male patients with these ABEL1 variants had genital abnormalities, in particular hypospadias. This is also something that some of the adult individuals in particular only told us after we asked specifically if there were any abnormalities about their genitalia. So it's one of the old lessons that we all learn in medical school that, you know, sometimes you will only hear about symptoms if you ask for them very specifically. And with that, I will hand over the control to Dr. Chen, who will tell us about the functional assays, the in vitro assays that were done for the ABEL2 variants that I just presented. Okay, so the interpretation of novel missense variants can sometimes be tricky. I mean, some of the variants are benign and some of them have deleterious effects on the gene or on the gene product. So according to ACMG standards and guidelines, the impact of the missense change depends on criteria such as evolutionary conservation of an amino acid for nucleotide, the location and context within the protein sequence, and the biochemical consequence of the amino acid substitution. So in this study, both amino acid residues are highly conserved from human to zebrafish. And both variants are absent in population database, including EXACT and NOMAD. And both variants are predicted to be disease causing or damaging by different in silico tools, including SIFT, polyphenol-2, and mutation taster. So ABL-1 has two isoforms, 1A and 1B, resulting from an alternative splicing of the first exon. And isoform 1B has 19 additional amino acid residues at the N-terminal. And the nearest correlation of these residues plays a role in the auto-inefficient of ABL-1 kinase activity. And in our study, the two missense variants are expected to affect both isoforms. ABL-1 has multiple functional domains. And tyrosine-245 is located within the linker region between SH-2 and kinase domain. While alanine-356 is located in the kinase domain. Previous study shows that both amino acid residues are the key player in regulating ABL-1 kinase activity. Tyrosine-245 is required for the activation of ABL-1 kinase activity by auto phosphorylation. And the other variant, alanine-356, would bind to the N-terminal nearest correlation group and form an auto-inhibitory conformation. So to investigate the potential effects of the two ABL-1 variants, we transiently transfect HEX cells with the plasmids expressing either wild-type or ABL-1 variant. All of them carry the B5 epitope pack. And then we perform immunoblot analysis to test for overall tyrosine phosphorylation and also the ABL-1-specific substrate as an indication of ABL-1 kinase activity. So starting from the left is the MT vector control followed by wild-type ABL-1, and then the two variants, tyrosine-245-16 and alanine-356-bran. So with the similar transfection efficiency, we show that there's increased phosphorylation of overall tyrosine. There's, I mean, overall tyrosine phosphorylation in both ABL-1 variants compared with wild-type ABL-1. So, I mean, I'm talking about this one, the overall phospho-tyrosine level. There's increase in the variants compared with the wild-type ABL-1. And also we see an increase phosphorylation of ABL-1-specific substrate that includes ABL-1 itself and step five. As for the phosphorylation of other proteins such as CRKL and SMAD2 and 3, we did not see a huge difference between the variants and the wild-type ABL-1. And I would like to mention that the reason that you do not see an ABL-1 autophosphorylation in tyrosine 245 cysteine is because the antibody we use specifically recognize phospho-tyrosine 245. So overall, we show that both variants increase ABL-1 kinase activity. And actually using the same functional study, we and our collaborators identify three additional de novo novel variants in the patients who shares overlapping clinical features. So in summary of this study, we show that germline mutations in ABL-1 cause a novel human developmental disorder. And these mutations cause increased ABL-1 kinase activity. And the disorder is characterized clinically by congenital heart defect, skeletal malformations, and failure to thrive. And the study was published last year. So for any one of you are interested and wanted to know more details, you can search this article on the PubMed. And next, Dr. Wang will give you another example of novel disease gene discovery. Thank you. Hello, everyone. This is Xiao Wang from Baylor College of Medicine. Thanks for attending the webinar and thanks Chuan and Christian for the prior talk. So I will continue with the second example. So the second study starts from a eight-year-old male. He had cystic hygroma and intrauterine growth restriction. He had a global developmental delay and regression at about four years old. He had diagnosis of severe autism and some seizures. His brain MRI found prominent anterior and posterior horns of left lateral ventricle. He had dysmorphic features, including multiple hair rolls, pertosis, and epicanthal folds. He also had clonodactyly and other digital anomalies. Prior workup, including chromosomal microarray study and sequencing analysis for genes associated with intellectual disability and congenital disorder of gycosylation were unrevealing. Then the patient sample was sent for whole exome sequencing at Baylor. Whole exome sequencing identified a de novo mosaic missense change in TRAF7 gene. This variant is de novo, not present in the parent. At that time, it was not present in any public databases, such as NOMAD and EXACT. And it is predicted to be deleterious by multiple prediction algorithms, including SIFT, polyphen, and mutation taster. So at that time, missense variants in TRAF7 have only been reported in two patients with autism, but without detailed clinical phenotypes. So I find out this clinical report, and I find all these variants as a variant of unknown significance. And I think it is related to the phenotype in our patient. And I think the TRAF7 gene is a candidate disease gene. Notably, this is a mosaic change. So by whole exome sequencing, we found 14 mutant reads and 105 total reads supporting this variant, covering this variant. So the allele fraction or the level of mosaicism is roughly about 13%. Later, we used PCR to amplify this variant from the patient sample, and then followed by high-throughput sequencing at a much higher depth of coverage. And the allele fraction is about 15%. So I'm showing you the Sanger data that we got for this trio. In the middle, the... As you can see, this blue peak is the wild-type peak, and the much lower black G peak is the mutant signal for these mutations in the poor man. And usually, if it's a heterozygous, the two peaks should have the same height, but the mutant peak here is much lower, suggesting it's a mosaic change. And then we use a restriction enzyme to digest the wild-type DNA and to enrich the mutant DNA and, as you can see, the mutant peak goes back, likely similar to that level of the wild-type peak. So just for the support that this is a mosaic change. So in total, we have recruited seven patients in this study. We identified four unique missense changes in TRAF7 from the seven patients. That means there is a one change that is recurrent in four subjects, which is listed in this table. It is argin-655 glutamine. In six out of the seven subjects, the variants were found to be denoble. In subject four, the variant is negative in the mother while the paternal sample was not available, but it is this recurrent variant. So still, it is a very good candidate variant. And for these four missense changes, they are absent in NOMAD and Xact databases. They all affect the amino acids that are highly conserved. Most of the prediction algorithms predict these variants to be deleterious. So in the first figure, we showed the cross-species alignments of the four amino acids that are affected by the changes we identified. And as you can see, they're conservative from human to zebrafish. And panel B is the exon-infron structure of TRAF7 gene. TRAF7 has 21 exons. And as you can see, the four variants are localized in the three prime exons. And panel C are the PERTN domain structure for TRAF7 PERTN. TRAF7 PERTN has four important domains from N-terminal to N-terminal. From N-terminal to C-terminal is RAIN domain, the zinc finger domain, coil-coil domain, and the WD40 domain. Two out of the four variants are localized in the coil-coil domain, and the other two variants are localized in the WD40 domain. And previous studies suggest that the WD40 domain is important domain for binding of the TRAF7 partners. So just to summarize the key clinical features of individuals with possibly disease-causing variants in TRAF7 gene, for the individuals that we were able to evaluate their developmental milestone, they all have developmental delay. Most individuals had congenital cardiac malformations and they're quite variable from ASD, VSD, PDA, and to other structural changes. Most of these patients have structural brain malformations and they're also quite variable. On all of the patients had digital anomalies such as clinodactyly and overlapping toes. And they all share dysmorphic facial features. So here I'm showing you some dysmorphic features from representative patients. The shared dysmorphic features include epicanthal folds, pertosis, abnormally set or dysplastic gears, low hairline, excess nuchal skin, and multiple hair rolls. Also in panel C and F, you may appreciate the overlapping toes. And interestingly, RAS opus C was suspected for four subjects in this study. So in the subject one had been evaluated for possible diagnosis of cardiofacial cutaneous syndrome and was negative for Newnan gene sequencing analysis. Subject two was suspected of having Newnan spectrum condition and was negative for a Newnan panel testing. Subject three has features reminiscent of Karst-Dillow syndrome, including a coarse facial appearance, congenital heart defects, and neuromuscular involvement, including scoliosis and contractures. And there is one unpublished subject who has possible disease-causing variants in TRAS7 and was suspected to have Newnan syndrome and was negative for Newnan gene testing. With that, I will turn the slides to Dr. Chen and he will further talk about how we performed a functional study on the variants that we identified in the TRAS7 gene in this patient. So TRAS7 is a tumor necrosis vector receptor-associated vector 7. It is an intracellular adapter that binds TNF receptors and transduce the cellular effects, which is mediated by TNF ligands. And TRAS7 is known to interact with subfamilies of mitogen-activated protein kinase, that's MAP kinase, and including ERK1 and ERK2, C-June NH2 terminal kinase, that's JNK, and P38. This is oversimplified pathway of what I just mentioned. So upon TNF ligand activation, TRAS7 will interact with MAP3K3 through the WD4T domain. And in cell culture system, co-expression of TRAS7 and MAP3K3 will lead to the activation of P38, JNK. Yes, and on the other hand, ERK1 and ERK2, the signaling cascade, have not been directly linked to TRAS7. So what we know is that ERK1 and ERK2 can be phosphorylated and activated by MAP2K1, which can in turn be activated and by phosphorylation by multiple protein kinase, including MAP3K3. So, sorry. So to investigate the potential effects of TRAS7 variants, similarly, we transiently transfect HEK cells with plasmids expressing either wild type or TRAS7 variants, and then followed by TNF alpha treatment in a cell culture. The reason we treat the cells with TNF alpha is because we would like to compare the TRAS7 variant effects on MAP kinase signaling, both in the baseline and in the activated state. So starting from the left is the vector, empty vector control, followed by a wild type TRAS7 and the four variants. And we see that expression of TRAS7 does not lead to change in overall ERK1 and ERK2 protein levels, as shown here. However, we do see that there's decreased phosphoERK1 and phosphoERK2 in TRAS7 variants, compared with the wild type TRAS7. And the reduction of phosphoERK1 and ERK2 is observed both in the activated state and in the baseline state. In other words, we see the similar phosphoERK1 and ERK2 phosphorylation reduction with or without TNF alpha treatment. And we also check what's the effect of TRAS7 variants on the other two MAP kinase signaling pathway, that is P38 and junk. However, we did not see a huge difference between the TRAS7 variants and the wild type. And here just shows you the quantification of phosphoERK1 and ERK2 in those transfected cell lines. So in summary of this study, we show that mutations in TRAS7 cause a novel human developmental disorder characterized by developmental delay, congenital heart defects, digital anomalies, and facial dysmorphism. And the in vitro assay shows that pathogenic missense variants associated with this disorder cause reduced phosphoERK1 and ERK2. So the study was published about two months ago. So for anyone that is interested and want to know more, please search on the PubMed. And with this, I'm going to give it back to Dr. Sha. Thank you, Dr. Chen, and thank you, Dr. Wang. So what we showed you using the two examples of AVIL1 and TRAS7 is how to move from an individual patient encountered in clinic to something that is publishable. And I would make that statement that, obviously, it has become much more difficult and much more is required today in order to publish those cases when comparing this to a situation three or four years ago. So very clearly, a single case, a single de novo variant identified in a novel not yet known to be disease-causing gene will be extremely hard to publish in any journal. So the step going from one individual patient to a series of patients is really critical in order to get it published. This is the genetic evidence that really weighs very heavily, and this is where you depend on either databases at your own institution, public databases of affected individuals receiving and undergoing genomic testing or initiatives such as GeneMatcher. What I want to show you in this last part of the publication is how do we determine strong evidence for a true gene-disease relationship. And that kind of goes beyond the initial publication. This is really for the community. When do we as a community feel very strongly that there is sufficient evidence to say that variants in a particular gene cause a bona fide human genetic disease? And I think that step requires much more than just a case series of five patients and much more than the initial publication. So I want to tell you, for those in particular who are not familiar with ClinGen yet, I want to tell you a little bit about ClinGen, a coordinated effort around the world where physicians, molecular geneticists, genetic researchers, genetic counselors, and bio-curators come together to work very hard to establish a database that provides us with the information as to how much evidence there really is for gene-disease associations and for variant-disease associations. So there are two different steps that we need to discriminate, and it's important to understand that these are two separate steps in the curation. First, teams of curators will look at the level of evidence that exists to determine whether a gene is related to a human disorder and whether variants in that particular gene provide enough evidence to suggest that mutations in this gene will truly cause a bona fide human disease. Once that is established, then the teams may move to variant curation, which goes into more detail and looking at single variants and specific variants to determine whether those very specific variants are sufficient to cause the established phenotype. This is a lot of hard work, and I can tell you that from first-hand experience, as I am one of the co-directors of the ClinGen workgroup for intellectual disability and autism spectrum disorder, it's really a daunting effort. It's a combination of the hundreds, if not thousands, of genes that are probably related and variants of which are causing intellectual impairment and or autism spectrum disorder. I just want to walk you through the different types and different levels of evidence that we assess and curate in order to determine whether a gene is truly associated with a human disease. There are very well-established standard operating procedures that the curators go through, and they will determine evidence looking at different types of information. So first is the genetic information, and you can see this here. This is the case-level data. So this is where the curators go back in PubMed and pull all publications that have reported variants in a particular gene that may cause a specific human disease. Now you realize that there are also human disease genes, mutations in which may cause different kinds of disorders, right? So different kinds of variants in one particular gene may account for different types of disorders. That's where the pre-curation comes in, and the team needs to decide which specific disorder they're looking at. But for example, if we say ABLE1 that we presented and we want to look at the evidence, then we would say, okay, this is the disorder that has been described that is characterized by congenital heart defects and skeletal malformations, and we want to see how much evidence is there that ABLE1 causes this particular condition. So there's the variant evidence that will be looked at, and this depends on how many individuals from how many families have been reported. It takes into account whether the variants that have been reported are de novo or not. If they are not de novo, then the team will look at segregation evidence and will look to see if the particular variant segregates with the phenotype within the family. In addition, there are different levels of evidence, different point levels that are given depending on whether the variant is a null variant. So if it's proven to be a null variant, there are higher points than if it's just a missense variant. And if it's a missense variant, the team will take into consideration whether there's functional evidence to show in an in vitro system that the missense variant actually has functional consequences within that system that may be relevant to the disease. In addition, case control data are taken into consideration, so this is looking at large databases. This is particularly relevant when you think about, let's say, intellectual disability or autism. These are cohorts where thousands of individuals or hundreds of individuals with a particular phenotype have been sequenced, looking at whether a certain variant is overrepresented within that cohort when compared to a control cohort. Beyond the genetic evidence, we will be looking at functional and experimental evidence. So there are additional points, and you can see the points listed here as a maximum score in the back, that are given for functional evidence, looking at biochemical function, including interaction data, expression data, whether the protein is specifically expressed in a tissue that is most relevant to the phenotype, functional alterations, models, non-human model organisms, cell culture models, and whether in those model organisms rescue has been shown, so that if you reintroduce the wild type protein, whether the phenotypes that have been reported can be rescued. So all of this will be tabulated and will give a total score. So we will include genetic evidence data, experimental evidence, and with that, we will get to a total number of points, and depending on the number of points, we will determine whether there's limited, moderate, strong, or definitive evidence for this gene to be a human disease gene. It's important to also note that in order to reach the definitive status, we do not only need a large number of points from evidence of genetic evidence and experimental evidence, but also we need replication over time. So we will look at whether the data that we have are only based on a single publication, and that would actually not be sufficient. So if we have just a single publication of 10 patients plus experimental evidence, but there hasn't been any replication by separate groups over time, we would not give that gene-disease relationship a definitive status. So this is a very labor-intensive process, but I think it's very important for the community as we are looking very carefully whether particular genes are related to human genetic disease. If you are interested in becoming a member of such a curation group, you can go to the ClinGen website, and there is a volunteer button, and you can sign up and express interest in becoming a member of a curation group within ClinGen. In summary, we showed you that there are still many novel human disease genes to be discovered. We showed you, using the ABLE-1 and the TREF-7 examples, that missense variants are generally more challenging to be interpreted, and we feel that functional assays are needed in order to determine whether those missense changes are really relevant or not. This is also reflected in the ClinGen SOP, as missense changes get less points than loss-of-function variants. And ClinGen represents one large, coordinated, structured effort to determine the level of evidence for gene-disease relationships. These are our references. We already mentioned the publications in Nature Genetics and the American Journal of Human Genetics that we talked about today, and you can have access to the ClinGen standard operating procedures. These are publicly available on the ClinGen website, which is www.clinicalgenome.gov. And with that, we will move to your questions and answers. Just, you know, to repeat, if you want to ask us a question, please type your question into the Q&A box in the lower left corner of your screen. And we have already received several interesting questions. Let me go into those. Ina Amarillo from Washington University in St. Louis has asked two questions about ABL-1. First, she was wondering if functional studies were performed in CML cell lines on a CML background, cells that have a BCR-ABL fusion, and those studies have not been done. The functional studies that we presented to you today were based on an immortalized cell line after ABL-1 transfection or transfection of the mutated constructs of ABL-1. Ina Amarillo also asks, what do you do if you have a de novo ABL-1 mutation? And if you do not have access to large databases, should you report it? Should you dismiss it or ignore it? I'm not sure if Dr. Amarillo refers to the situation now after our initial publication on these ABL-1 variants. If you're referring to that situation, I would encourage you to contact us as Dr. Wang continues to collect cases, and we've been contacted by several physicians and research groups over the past year who have additional cases of individuals with de novo variants in ABL-1. If you were referring to a theoretical situation saying that you have a de novo variant in a gene that is not yet known to be a diseased gene, what do you do with it if you do not have access to large databases? In that case, I feel that situation is much more challenging. You may contact large groups for whom you know that they've done large numbers of clinical exome sequencing, and you may reach out to them and bring that gene to their attention and ask if they have more cases and if they want to collaborate with you, or you may want to go to GeneMatcher and see if you match for your gene. But certainly, if you only have a single case in a potentially novel diseased gene, that will be very difficult these days to get that story published. Now Elena Repnikova from Children's Mercy Hospital also has a question about ABL-1. She feels that we have provided evidence of increased expression of the protein in an immortalized cell culture, but how do we know that the phenotype is indeed due to increased expression of ABL-1? She says, you know, we should study the missense changes of ABL-1 in mortal organisms, for example mice, and then see if they recapitulate our phenotypes. So first of all, you know, let me say that we were lucky not to have Dr. Repnikova as our reviewer for our Nature Genetics paper, but, you know, joking aside, you are right. We did our functional assessment in an immortalized cell line, and it is based on transfection and overexpression of ABL-1 constructs. I feel that the major evidence for our study came from the genetics data, showing that de novo variants in ABL-1, and not only that, but also in many of those cases, the same exact de novo variant in ABL-1, which was absent from control databases, occurred in individuals who share clinical phenotypes. I feel that is really the strong evidence that there is. The functional data in the immortalized cell lines was not necessarily to prove pathogenicity, but to look and find out in which direction the changes are and whether the changes lead to an increased activity of ABL-1 as it relates to phosphorylation in the cell or decreased activity. But you're absolutely right that, you know, this may be in part artificial and that it should be proven, A, in different kind of cell lines, which actually we did. But it would certainly be of great interest to generate transgenic humanized mice who have the same missense variants that we found in our patients with ABL-1 variants. That we have not done. I can tell you that ABL-1 knockout mice exist, and they do have congenital malformations, some of which overlap with what we see in the patients. And also, it's of interest to us to see literature about Gleevec, the selective inhibitor of BCR-ABL. And when Gleevec has been given to women with CML who were pregnant at the same time, there seems to be an increased chance of them giving birth to children with congenital malformations, some of which, again, overlap with the features that we see in our patients. That may suggest, even though it is certainly not definitive evidence, but it may suggest that the level of ABL-1 expression and function is critical at certain developmental time points. And if it is altered, it may cause or contribute to congenital malformations. Shujuan Pan from Myriad Genetics asks if all of the TRAF7 patients wear mosaic, and if yes, do we think that it is lethal if it's heterozygous? And I think Dr. Wang can answer that question. Okay. Actually, only one out of the seven patients had the mosaic change, at least based on the observed allele fraction calculated from the whole exome sequencing data. So all the other six patients, they appear to be heterozygous for their mutation. And then Dr. Pan from Myriad Genetics also asks if there is a reference for the method used to confirm mosaicism using enzyme digestion. So basically what we did is that we found a restriction enzyme that can specifically digest the wild-type sequence, but not the mutant sequence of the DNA. And after, you know, we order the enzyme and then digest the DNA for a certain amount of time. And after that, we use the PCR and do sequencing. So there's not a reference, but the method is relatively straightforward. And we have one question from Charles Saley from MTL asking, only doing functional studies might not be enough to show a true causal relationship of the gene. Did you do anything further than just the functional studies? I think the answer to this is no. We only collected genetic evidence and then supplemented that with some functional studies to show that there are functional consequences of the missense changes. But you're absolutely right that, you know, this is not definitive evidence for the causal relationship of the gene with the disease. In ClinGen, we would consider the functional evidence as part of our curation and it would get us into a higher score for the missense variants when compared to a situation where only the de novo missense variants are reported without any functional evidence that the missense change actually causes a functional change within the cellular system. But it is certainly not enough to determine a definitive relationship. This is Xiaowang. Maybe I can add a few comments. We do have, you know, additional replicated cases. So, you know, after the publication of ABL-1 and TRAF7-STARI, for example, for ABL-1 now, we have additional 5-plus patients. They have de novo mutations in ABL-1 and they share similar features to the reported ones. Even some mutations turn out to be like a hotspot, mutation hotspot, very recurrent in several more additional patients. The same thing for TRAF7, there is a recurrent variant and we have seen it in one additional patient and we have a few other patients that have de novo mutations in this gene. Thank you. I think with that, we have answered all the questions that have been asked, so let me just remind you that the next ACMG Annual Meeting is coming up soon. This is April 2nd to 6th of next year in Seattle and the Call for Abstracts will open soon. The submission site will be open in October of 2018 and the deadline for submission of your abstracts is November 30th, 2018. So please keep that in mind as you have interesting stories that you want to tell and show. We want your best research at ACMG 2019, so we look forward to your submissions. You can submit those on the ACMG website, the meeting website, www.acmgmeeting.net. With that, we are coming to an end. I want to thank you for attending our webinar and I want to thank again QIAGEN Bioinformatics and the Ingenuity Clinical Decision Support Platform for their support of this program. A recording of this seminar will be available in the ACMG Genetics Academy very soon. I just want to remind you that there are handouts that have been loaded under conference materials that you can have access to before you exit the webinar. Thank you again and hope to see you again soon.

ACMG Genomics Case Conference

disease gene discoveries

genetics

functional evidence

Baylor College of Medicine

QIAGEN Bioinformatics

Ingenuity Clinical Decision Support Platform

ABL1 gene

TRAF7 gene