Khurana: (For details on this ENCODE work see Gerstein el al Nature 2012 PubMed: http://t.co/nhVZbo7KdP ) #AACR14
1:00pm April 5th 2014 via Hootsuite
Khurana: Looking at TF binding regions, assign to target genes, correlate to exp. data. 119 TF's, 9K target genes, 28K edges #AACR14
12:59pm April 5th 2014 via Hootsuite
.@teamoncology It is known that the gut absorbs many types of molecules, first time I heard that microRNAs are included in that
12:57pm April 5th 2014 via Hootsuite in reply to
.@teamoncology Looking at RNA-Seq from plasma, ID'd microRNA from corn (from a Western diet !); rice miRNA from an Asian one. Fascinating.
12:56pm April 5th 2014 via Hootsuite in reply to
Khurana: Figure from Ecker Nature 2012 PubMed http://t.co/K3elDXC3va showing chromatin organization, functional elements #AACR14
12:54pm April 5th 2014 via Hootsuite
Khurana: Nice illus. of data: TCGA 2.5PBytes, TCGA 910 TBases in CGHub; TGB 220GB; ADSP 46GB #AACR14
12:52pm April 5th 2014 via Hootsuite
Khurana: ncRNA from WGS; from TCGA, TGP, ICGC, ENCODE consortia. From cancer genomes, to population scale seq, to non-coding #AACR14
12:51pm April 5th 2014 via Hootsuite
Next up: Ekta Khurana Yale Univ "Information in non-coding DNA" #AACR14
12:49pm April 5th 2014 via Hootsuite
Guo: Illustrated exponential growth of publications of WES / RNA-Seq, possibilities for data mining #AACR14
Guo: 'Food RNA' - food microRNA in human samples. Diff. rice / corn in diet Wang PLOS One 2012 PubMed http://t.co/rzvdurCABD #AACR14
12:48pm April 5th 2014 via Hootsuite
Guo: Mining RNA-Seq data, can look at SNV & Indels; even microsat instability. Can look at RNA editing, allele-spec expression #AACR14
12:43pm April 5th 2014 via Hootsuite
Guo: Also looking at mtRNA from WES, and also viral sequence (HBV, HPV) #AACR14
12:41pm April 5th 2014 via Hootsuite
Guo: Took Tibetan WES of 50 individuals, and published non-exome high-quality SNVs. #AACR14
12:38pm April 5th 2014 via Hootsuite
Guo: Since capture isn't perfect, you get a lot of add'l data (off-target); ¬50% off-target; intron / intergenic functions #AACR14
12:34pm April 5th 2014 via Hootsuite
Guo: Known knowns, known unknowns and unknown unknowns: we know SNV, CNVs, structural var's. Others: non-targeted SNV, mtDNA, add'l #AACR14
12:32pm April 5th 2014 via Hootsuite
Next up: Yan Guo Vanderbilt Univ "Lost treasures in sequence data" #AACR14
12:30pm April 5th 2014 via Hootsuite
Reis-Filho's comment: Mentions E. Mardis could help on the crowdsourcing effort #AACR14
12:29pm April 5th 2014 via Hootsuite
Reis-Filho: (HER2 comment): Could get a nice repository of knowledge. #AACR14
12:28pm April 5th 2014 via Hootsuite
Reis-Filho: (HER2 comment): Rapidly exceeds any cancer biologist's expertise; suggest using Wikipedia (seriously) on genes #AACR14
12:27pm April 5th 2014 via Hootsuite
Reis-Filho:Q (comment from person who did HER2 work): review from structural biologist yields best predictions (not scalable) #AACR14
Reis-Filho: Concl: drivers vs passengers still an inexact science; need data from multiple levels #AACR14
12:26pm April 5th 2014 via Hootsuite
Reis-Filho: In functional validation though, the story may not be so clear-cut #AACR14
12:23pm April 5th 2014 via Hootsuite
Reis-Filho: Functional studies 9/14 were drivers functionally. I767M in HER2 called driver by several callers, but not functional. #AACR14
12:22pm April 5th 2014 via Hootsuite
Reis-Filho: No predictor called all the 14 mutations in HER2 #AACR14
12:21pm April 5th 2014 via Hootsuite
Reis-Filho: Looking at BRAF and PIK3CA specific mutation constructs, these approaches work. But - combining predictive + functional? #AACR14
12:20pm April 5th 2014 via Hootsuite
.@drdonsdizon You are most welcome - enjoying these talks immensely (just taking notes and sharing them!)
12:19pm April 5th 2014 via Hootsuite in reply to
Reis-Filho: MutationAssessor with TCGA was only 74% acc, 89% sens. Better; but 'no perfect predictor available' #AACR14
12:17pm April 5th 2014 via Hootsuite
Reis-Filho: Results of the 'Pepsi Challenge': CHASM had 89% accuracy. But trained from COSMIC. W/ TCGA: 50% acc.! 0% sens. #AACR14
Reis-Filho: Gnad et al 2013 PubMed http://t.co/WH32ALbNxB looked at mutation predictors. Most popular is CHASM, leveraging COSMIC #AACR14
12:15pm April 5th 2014 via Hootsuite
Reis-Filho: These mutation predictors look at conservation, struct changes, prot annotations, training sets plus bioinf. #AACR14
12:14pm April 5th 2014 via Hootsuite
Reis-Filho: Looking at many mutation predictors (PolyPhen, SIFT et al) - they come out every week #AACR14
Reis-Filho: MuSiC reference for predicting driver mutations Dees et al 2012 PubMed: http://t.co/AMV56zJYhP #AACR14
12:13pm April 5th 2014 via Hootsuite
Reis-Filho: Predicting drivers: three papers as examples MuSiC, Two others (PubMed): #AACR14 http://t.co/JP52QvJhnM http://t.co/wdM1iqB4WC
12:12pm April 5th 2014 via Hootsuite
Reis-Filho: Passengers - no selective advantage, looking at Titan (TTN) - 16% of Br Ca have TTN mutations. TTN exp. in muscle #AACR14
12:09pm April 5th 2014 via Hootsuite
Reis-Filho: Tumor suppressor drivers are truncating or frameshift mutations #AACR14
12:08pm April 5th 2014 via Hootsuite
Reis-Filho: Other type of driver is tumor suppressor mutations - TCGA lollipop plot is even distribution (p53 in Br Ca) #AACR14
12:07pm April 5th 2014 via Hootsuite
Reis-Filho: One types of drivers: oncogenes (detected via hotspot or fusion genes via RNA-Seq); showed TCGA 'lollipop' freq plots #AACR14
Reis-Filho: Driver alteration defined: a net growth advantage, positively selected for in tumorigenesis #AACR14
12:05pm April 5th 2014 via Hootsuite
Reis-Filho: Showed Circos images - mutations he's never heard about, never described before, but may be of interest #AACR14
12:04pm April 5th 2014 via Hootsuite
Next up: Jorge Reis-Filho MSKCC: "Identification of driver versus passenger mutations " #AACR14
12:03pm April 5th 2014 via Hootsuite
Hatzis: "Treat", "SIMPLEX", "WEP" were three of the five samples of these integrated solutions #AACR14
12:01pm April 5th 2014 via Hootsuite
Hatzis: Other challenges: high bioinformatics barrier. But there are several integrated pipelines #AACR14
Hatzis: Challenges: NGS reads short: difficult to assemble / map; Bias; Ampl. bias (polymerase slippage in poly(A)); Errors #AACR14
12:00pm April 5th 2014 via Hootsuite
Hatzis: Also a figure to look at structural variant calling from Alkan 2011 PubMed: http://t.co/kpsuKFEG4U #AACR14
11:58am April 5th 2014 via Hootsuite
Hatzis: Showed briefly a chart of 8 different algorithms for var calling, from Nielsen et al (2011) PubMed http://t.co/btRciPyy8e #AACR14
11:56am April 5th 2014 via Hootsuite
Hatzis: A particular location may need a local assembly to solve alignment problems; outlined .BAM file format, getting to .VCF #AACR14
11:54am April 5th 2014 via Hootsuite
Hatzis: Illustrates the workflow overview for SNP calling pipeline from Altmann et al 2012 PubMed: http://t.co/VIEo7xUXyQ #AACR14
11:50am April 5th 2014 via Hootsuite
Hatzis: Best practice - GATK paper from the @BroadInstitute mentioned http://t.co/qdLMiVoq2Z
11:48am April 5th 2014 via Hootsuite
Hatzis: (DARPA's challenge was called 'all your shreds are belong to us') http://t.co/rnujGd8agT #AACR14
11:47am April 5th 2014 via Hootsuite
Hatzis: DARPA had a shredder puzzle challenge (10K pieces solved in 30d) in 2011 (fun with informatics) #AACR14
11:46am April 5th 2014 via Hootsuite
