 So, I'd like to echo the sentiments and thank the organizers for selecting this abstract. It's my great privilege to present the work of BC Cancer Agency in an exciting project I hope you'll agree in precision medicine or personalized oncogenomics. Just a little background about the BC Cancer Agency. The population of British Columbia is 4.5 million but live in an area about twice the size of California. We're in quite a unique setting in that we can offer a provincial population-wide cancer control program and this reaches from prevention, screening, diagnosis through treatment. The scope of the POG project, this personalized oncogenomics, is to bridge the gap really between genomics research and clinical practice at the center. We aim to identify tumor-specific therapeutic targets in cancer patients and in this POG phase one project with late-stage metastatic disease. So the process is outlined in the flow diagram. At the bottom begins with, of course, a patient consult and consent process. We typically collect tumor biopsies, we'll also request an archival FFPE block and match peripheral blood for that patient. And then we construct PCR-free whole genome libraries of the tumor and the peripheral blood normal. We'll also perform transcriptome sequencing as trans-specific RNA-seq. We sequence on the alumina high-seq instruments and then once a week we meet with the referring clinicians, discuss the relevant findings and of course submit a report for their attention with recommendations. This is a very busy slide and I don't intend you to take a lot from it other than the fact that this is, it is complex but it is also structured and the color coding indicates the sub-departments within our bioinformatics department that are analyzing this data. And I've highlighted just a few of the areas in which TCGA data is actively contributing towards this project. So these include the comparison to expression database and I'll show you a few examples of that in case studies, the variant calling, algorithms both from transcriptome and from the genome to help filter our calls for somatic mutations. And of course all of this feed from samples, feed through this process to a final report. The study has thus far enrolled 83 patients, eight of these are pediatric cases. We've performed 69 biopsies to date, representing 28 tumor types and the frequency of those tumor types are shown in the bar plot. You can see there's a bias towards a collection of breast cancer samples, primarily down to the very active participation of breast cancer clinicians within the BCCA. We generate over 90x coverage of the tumor genomes and almost 50x coverage of the archival and matched normal genomes and over 300 million reads on average for a transcriptome library. Importantly we've reported 50 cases and counting and the average time thus far is 38 days from biopsy to the report. So here are some of just four of the areas in which POG has guided treatment decision making. I won't have time to go over all but we'll give case studies for the top two. So providing directed cytotoxic chemotherapy and of course the targeted therapeutic options were available. And the second one on the list is the complimenting or even correcting FDA approved clinical tests. We also can identify cases of change diagnosis or provide a diagnosis where previously the primary tumor was unknown. So jumping to our first case study, we call POG 3, it was a squamous cell carcinoma of the skin. This gentleman arrived in 2007 having had a red rash on his upper chest. He had actually been living with this rash for well over a year maybe even two years and when he presented was had bleeding ulcerations. It was diagnosed with squamous cell carcinoma and he began multiple lines of chemotherapy and radiotherapy began in 2007 and persisted through to about 2012 when a new node near his right ear appeared, the pre-oracular node. Both this node and his chest mass were growing throughout the course of 2012 to a degree where the ear node was really preventing his hearing and causing extreme pain. So in September 2012 both the node and the chest lesion were biopsied for this study. The somatic mutation status of the tumors is illustrated in the Venn diagrams here and strikingly not only are there many somatic variants as one might expect from cancers that have been developing and subjected to a number of rounds of therapies. So they're in the high thousands but there was very little overlap between these two lesions. They looked at really very different cancers indeed and no overlap when we looked at the small indels. Likewise when we look at the copy number profiles for these two tumors the chest is shown in the circost plot in the outer ring and the pre-oracular node on the inner ring and you can see that there's really relatively little commonality between these two tumors at the copy number level. If we just highlight chromosome 16 and take a look at the copy number plots again the break point regions are few and far between. There are areas of gain at 16Q that are not apparently gained in the chest lesion loss including a homozygous loss on chromosome 16. And so we could put this data together in the form of pathway analyses and so this is integrating now the genome and the transcriptome data. And this is looking at the chest tumor again and you can see that there are many rearrangements consistent with DNA repair defects that were detected. But importantly and highlight here both loss of P10 both in copy number and decreased expression and concomitant increase in AKT expression and a gain of copy number. And so this was suggestive of treatment options. Contrast that with the pre-oracular tumor you can see now that we have very elevated EGFR amplicons so both in expression and again copy number amplifications. Just to name a few. But these ones suggest and I apologize for the screamers in the orders. These two suggested treatment options of course. And so in the pre-oracular tumor the over expression of EGFR suggested a lot and the P10 homozygous loss and AKT gain for the chest lesion suggested treatment with everolimus. The patient was treated initially with both agents but exhibited quite a lot of toxic effects and so was taken off everolimus initially. The ear tumor responded incredibly well to a lot nib and decreased from just after biopsy very angry and red inflamed area and the clinicians tell me that this is really dramatic progress. The involution of the tumor was really recovering after three short weeks and also his hearing recovered in that right in that right ear. The pre-oracular tumor did progress sadly and a second biopsy was performed and in this biopsy the analysis showed even further extreme amplification of EGFR. So now to 55 copies of the gene and extreme gene expression level as indicated by the red line. This is now compared to all of the TCGA data and was the most extreme express of this that we've identified across the whole of those TCGA 8,000 plus samples. Turning to the second case study to finish, this was a 68-year-old Asian male lifelong never smoker who was diagnosed with non-small-cell lung adenocarcinoma in January of 2013. He had the approved clinical tests for EGFR and ALC through break apart fish probes and these were negative in the clinic. Subsequently had multiple lines of radiation and chemotherapeutic therapy between March and July of 2013 but continued to progress with the disease. And so in August we took the same node out for biopsy. Almost immediately the data the data yelled the transcriptome fusion between EML4 exons 1 to 13 and ALC exons 20 to 29 a really pretty canonical fusion for non-small-cell lung cancer patients. And yet I just told you that the clinical test had been a negative with only 3% of the cells exhibiting break apart. Closer inspection of the sequence both of the genomic level and the transcriptome level revealed that not only did the inversion that gives rise to this gene fusion exist but a much larger inversion of chromosome 2 and subsequent insertion into the chromosome 12 locus had prevented one of the vice's fish probes from binding adequately and hence the negative result in the clinical assay. If we look at the expression level as a consequence of this oncogenic fusion we see that again it's an extreme this particular patient is an extreme expressive for ALC here on the right 99th percentile of all expression compared to the TCJ long adenocarcinoma data set and ROS1 which is a second target for chrysotinib again was a high express in the 94th centile and so this course did suggest that the chrysotinib should be administered happily the patient was just well enough to receive chrysotinib within a day of the report being given to the to the clinician and the tumor responded really dramatically with it so within less than three months the tumors that you can see highlighted with the red arrows had really began to shrink away and in fact we're not visible for this this lower lesion and so to summarize the the overall Pog results for this for this phase one study phase one of this study for each patient we've sequenced three or more genomes and a transcriptome 80 83 content of patients now actually with advanced cancer have been enrolled 74 biopsies attempted some of them have failed giving us 69 biopsy materials we have full data available for those 50 patients and importantly they've been clinically evaluated in 38 cases and going back to the clinician the referring clinicians and asking can they ascribe actionable or treatable information from this genomic and transcript transcriptome data they report that in 87% of those cases there was action or treatment possible and it was offered into to 55% of those 33 cases of course a number of patients as in a late-stage setting had died during or shortly after the analysis and what's next for the personalized oncogenomics well we've we've just had approval for 5,000 cases and we intend to achieve these 5,000 cases within the next five years this will take us from about one patient a week on average currently to greater than one patient per day in the coming years we'll continue in emphasis on genome and transcriptome sequencing but we will include a more elaborate anchor panel for rapid turnaround time and a first look at these of these tumors when they arrive and we will bring forward from late-stage disease to fresh diagnostic biopsies and of course it increase always aim to increase our speed and accuracy of the sequence analysis and the report generation and and importantly we'll be verifying actionable results and continue to verify actionable results in a clinical lab prior to the treatment of the patient and finally just leave me to thank that the large multi disciplinary team at the BC Cancer Agency expertly led by on the clinical side Janessa Laskin and on the research side Marco Mara I'm very grateful to all the colleagues and most in particular the the patients and their families and the BC Cancer Foundation for funding this project and I have had to race through this a little and so I'll be happy to see you at poster number 10 if you have specific questions we can't address in the time left here but thank you for his attention what if any are the plans to make these data public so we obviously we need to be very careful about the confidentiality of of these individuals and we will we certainly need to discuss this more in depth there's no immediate plans to release the state it just is a lost opportunity and we should try to figure out all this wonderful sequencing helping patients we need to then use it to help all the patients and learn more absolutely so I mean of course there are the publications there's one publication in press currently but I mean it's the mining yes sure yeah one question very simple so do you worry about a tumor heterogeneity when you analyze your data yes we do and and so so one of the one of the aspects idea I didn't touch on this is that we require tumor tumor content of of over 40% before we will progress with all this deep genome and transcriptome sequencing heterogeneity certainly plays a big part and and in the analysis we we we try and tease apart as much of that heterogeneity as possible but a lot of it comes down to the sequence depth we're able to achieve with deep sequencing so for the time of interest let us move to the next speaker so for TCGA we face the same data but we use a different algorithm to interpret the data this particular highlighted by mutation calling so recently the SCGC and TCGA organized the dream challenge so that I see the speak from Dr. Paul Boros from the OIRC OICR talk about the summary from the first round SCGC TCGA dream challenge about mutation calling