Gutekunst: Puts up figure from Garraway JCO 2013 and the wide heterogeneity. Lists out 9 flavors of @illumina instruments #NGDx15
12:42pm August 18th 2015 via Hootsuite
Gutekunst: Predisposition to cancer - can it be screened for earlier? Prognosis? Treatment response? Monitorinig? #NGDx15
12:40pm August 18th 2015 via Hootsuite
Chair of Companion Dx session points out this resource: http://t.co/bSDhUotrZC #NGDx15
12:39pm August 18th 2015 via Hootsuite
Karen Gutekunst (Illumina CA) "Universal testing for actionable genomic variants in cancer: a paradigm shift in precision oncology" #NGDx15
12:38pm August 18th 2015 via Hootsuite
Powers: Reviews changes to clinical recommendations based upon number of patients with clinical actionable above family history #NGDx15
12:22pm August 18th 2015 via Hootsuite
Powers: ~9% BRCA1/2 pos, about 4% carry non-BRCA pathogentic var's. A 40% increase in yield #NGDx15
12:19pm August 18th 2015 via Hootsuite
Powers: Showed 100% sens and 100% spec against >50 samples with known del/dups #NGDx15
12:18pm August 18th 2015 via Hootsuite
Powers: Listed genes with '100% no-call rate': DNAF2, INPP5E, MBL2, NF1, NRAS, PMS2 Need other methods Validated against 869 smpls #NGDx15
12:17pm August 18th 2015 via Hootsuite
Powers: Test algorithm for PMP22 dupe, CFTR Del exons 2-3. Larger CNV events but examine target-by-target. #NGDx15
12:15pm August 18th 2015 via Hootsuite
Powers:Going into validation - use a 29-gene hereditary panel (now up to 64 genes) Very hard to find samples NIBSC for Lynch #NGDx15
12:14pm August 18th 2015 via Hootsuite
Powers: 3-exon deletion in BRCA illustrated against normal samples. Split-read detection (method 2) - detecting breakpoints #NGDx15
12:13pm August 18th 2015 via Hootsuite
Powers: Thus they are able to assign a Q-score to CN. Per-target p-values for copy-normal according to a goodness-of-fit metric #NGDx15
12:12pm August 18th 2015 via Hootsuite
Powers: Each target assigned a ploidy, and confidence. No-calls are repeated or reported. Quality - expected spreads vs actual #NGDx15
12:11pm August 18th 2015 via Hootsuite
Powers: Shows BRCA1 del exon 19; CNVitae leverages other methods, custom tool, specifically for constitutional events. #NGDx15
12:10pm August 18th 2015 via Hootsuite
Powers: Read depth is highly variable. Reads that map to a given area - exon-resolution by CNVitae tool #NGDx15
12:09pm August 18th 2015 via Hootsuite
Powers: 3rd is paired-end analysis; 4th is de novo assembly. This isn't used - computationally complex. #NGDx15
12:07pm August 18th 2015 via Hootsuite
Powers:Del/Dup events: read-depth analysis, analogous to microarrays. Split-read analysis: ends map to different regions #NGDx15
Powers:For Del/Dup analysis: NGS is digital, quantitative, very high res. CNV is clinically important - 20% of BRCA, >70% CharcotMT #NGDx
12:06pm August 18th 2015 via Hootsuite
Powers: In their experience pull-down (hyb) working better than PCR-based for enrichment. Reviews alignment, SNV and indel calling #NGDx15
12:04pm August 18th 2015 via Hootsuite
Powers: (NB: Mentions his involvement with DoveTail genomics.) 4-5% meet NCCN criteria are BRCA1/2 neg but carry other genes #NGDx15
12:03pm August 18th 2015 via Hootsuite
Martin Powers (In Vitae, CA) "Strategies for the validation of NGS-based multi-gene panels in cancer dx including copy number" #NGDx15
12:01pm August 18th 2015 via Hootsuite
Lebo: Lists many software tools available for annotating variants. #NGDx15
11:58am August 18th 2015 via Hootsuite
Lebo: Causative variant-finding is needle-in-haystack effort, depending on accuracy and completeness of annotations #NGDx15
11:57am August 18th 2015 via Hootsuite
Lebo: Validation of pipeline for SNVs very high sens (>99%), PPV 99.8%. Their medical exome had only 11 FPs #NGDx15
11:56am August 18th 2015 via Hootsuite
Lebo: Their sources - mentions same resources from GIAB (GeT-RM, GCAT, GIAB, NA12878) #NGDx15
11:52am August 18th 2015 via Hootsuite
Lebo: Can separate technical assay validation vs. bifx. Technical - inter- and intra-run variability. For bifx, coverage, annotation #NGDx15
11:51am August 18th 2015 via Hootsuite
Lebo: Validation - divided into 3 tiers: first - sens, spec, PPV, reprod, repeat, front-to-end. 2nd: sens, PPV 3rd: expected vs obs #NGDx15
11:49am August 18th 2015 via Hootsuite
Lebo: Commercial exome and supplemented with targeted - 96% of exons at >20x. By gene, >98% at >20x #NGDx15
11:47am August 18th 2015 via Hootsuite
Lebo: Working with Emory and CHOP, an enhanced exome w/4631 genes. Exon-level coverage bar chart - comparing RefSeq, Med Exome, ACMG #NGDx15
11:45am August 18th 2015 via Hootsuite
Lebo: They focus on RefSeq; also 3 capture kits all compared @illumina, @agilent and @iontorrent 29.6MB in common but wide swings #NGDx15
11:44am August 18th 2015 via Hootsuite
Lebo: Exome sequencing: no clear def'n. 5', 3' UTR? microRNA? unannotated transcripts? Typical - RefSeq or Ensembl. CCDS... #NGDx15
11:43am August 18th 2015 via Hootsuite
Lebo: For dilated cardiomyopathy (DCM) chart of genetic heterogeneity. ID'd new genes - benefit of large panel approach #NGDx15
11:42am August 18th 2015 via Hootsuite
Lebo: Listed cardiomyopathy panel, all add'l phenotypes (~40). Showed 5, 10 19, 24 and 46 genes, and neg/pos/inconclusive cases #NGDx15
11:41am August 18th 2015 via Hootsuite
Lebo: Able to detect single-exon gain <3 is very difficult - lots of FPs. Exome CNV calling - will need pool of normal samples #NGDx15
11:39am August 18th 2015 via Hootsuite
Lebo: Diagram of read depth from 10 samples over MYH7 gene: can look at dip in coverage across samples. VisCap a normalization tool #NGDx15
11:38am August 18th 2015 via Hootsuite
Lebo: Creation of panels - ones with homology issues. 'Many other clinically relevant genes have issues with current HTP seq tech' #NGDx15
11:37am August 18th 2015 via Hootsuite
Lebo: 20.8K common, but many 100's are unique in the Venn diagram. Levels of coverage include at base level, exon-level, gene-level #NGDx15
11:34am August 18th 2015 via Hootsuite
Lebo: Other var's require other tech- mosaicism, structural variations, homologous regions. Same sample WGS, WES1, WES2 pipelines #NGDx15
11:33am August 18th 2015 via Hootsuite
Lebo: Refers to ACMG clinical laboratory standards 2013 http://t.co/HKt142t2ZU Reviews their NGS analysis pipeline CNVs 'tricky' #NGDx15
11:32am August 18th 2015 via Hootsuite
Matthew Lebo (Partners MA) "From gene panels to exome sequencing: validation and implementation of NGS assay in the clinic" #NGDx15
11:29am August 18th 2015 via Hootsuite
Zook:A2: @PacBio calls include epigenetic calls, but the data is there. #NGDx15
11:28am August 18th 2015 via Hootsuite
Zook:Q:Is confidence binary? A:Yes. Most people only use high-confidence calls Q:Any interest in epigenome in a bottle? A:Not soon #NGDx15
Zook: Can sign up for updates at their site http://t.co/rfHI7pZvld GCAT exome comparison tool http://t.co/wTTbobhKmj #NGDx15
11:27am August 18th 2015 via Hootsuite
Zook: Francisco de la Vega, @mason_lab, Tina Graves, Valerie Schneider leading effort. FTP collecting data, other volunteers needed #NGDx15
11:25am August 18th 2015 via Hootsuite
Zook: New GIAB samples - have PGP Trio from Illumina PE, MP and Moleculo; CGI, CGI LFR; Proton Exome; BioNano Optical Maps; PacBio #NGDx15
11:24am August 18th 2015 via Hootsuite
Zook: Limitations of read-length for repetitive regions; other examples of FP/FN are complex vars, compd hets, regions w/many vars #NGDx15
11:21am August 18th 2015 via Hootsuite
Zook: 60-70% GC has higher FP rate; not clear why. Tandem repeats <7 also higher FP rate. Number of indels not assessed: similar #NGDx15
11:20am August 18th 2015 via Hootsuite
Zook: Works w/Global Alliance @GA4GH to benchmark, stratifying by 'genome context' - plot of high and low GC content vs confidence #NGDx15
11:17am August 18th 2015 via Hootsuite
Zook: One paper from BMC Genomics illustrating use of reference material http://t.co/4ufEjHr5go #NGDx15
11:15am August 18th 2015 via Hootsuite
Zook: Coriell news link about reference materials. http://t.co/h3098MOLB9 Reviews validation process with red and yellow-light steps #NGDx15
11:13am August 18th 2015 via Hootsuite
