April 2015

IG: For SIM, 16 submissions "no one could agree w/ each other". Established Gold set. Selected 5K positions, 2 HaloPlex Ion & MiSeq #AAC

11:42am April 21st 2015 via Hootsuite

IG: Each had to align & call: chart of SSMs: only 70 everyone agreed. Minus 1, 2 3: not good agreement betw. each other. #AACR15

11:40am April 21st 2015 via Hootsuite

IG: Provided everyone 40x tumor, normal from 184 samples. Asked for calls w/confidence. All same FASTQ, 20 submissions #AACR15

11:39am April 21st 2015 via Hootsuite

IG: Today looking at WGS. Nomenclature: SSM (single-base mut), SIM (somatic indel), SNV (germline var) SNP (where population AF >1%) #AAC

11:38am April 21st 2015 via Hootsuite

Ivo Gut (Centre Nacional d'Anàlisi Genòmica, Barcelona) "Experience of the ICGC somatic mutation calling benchmark" IG #AACR15

11:37am April 21st 2015 via Hootsuite

PB: Tumor het challenge: 50 single-sample datasets, cloud-only; also single individ. multi-sample datasets. #AACR15

11:31am April 21st 2015 via Hootsuite

PB: First paper in press Ewing Nature Methods. Next: two add'l challenges, one on RNA-Seq, other on tumor heterogeneity #AACR15

11:29am April 21st 2015 via Hootsuite

PB: Mutations profiles 'biased in all sorts of ways'. Sequence context - trinucleotide context. FP's across tumors: C>T deamination #AACR

11:28am April 21st 2015 via Hootsuite

PB: May be more efficacious - do a small set of validations and then tune for improvements. Of 16 teams: 20% improved scores #AACR15

11:27am April 21st 2015 via Hootsuite

PB: Challenge has >300 participants; coding regions highly accurate; recurring FP's in 40-50% of teams #AACR15

11:24am April 21st 2015 via Hootsuite

PB: Range of variables - between tumor/normal datasets, can model library quality with a fraction (1/8) the data Called SeqControl #AACR15

11:23am April 21st 2015 via Hootsuite

PB: The need for process control; difference in library quality (due to # of duplicates removed). Can it be predicted? #AACR15

11:18am April 21st 2015 via Hootsuite

PB: Huge discordance in br ca analysis '13 Genome Med ref http://t.co/LWk9Vf3NKy #AACR15

11:15am April 21st 2015 via Hootsuite

PB: Problem is that with different 'black boxes you get different results' (in terms of analysis of many different types) #AACR15

11:14am April 21st 2015 via Hootsuite

Paul Boutros (Ontario Inst Cancer Res Canada) “The ICGC-TCGA DREAM somatic mutation calling challenge” PB #AACR15

11:13am April 21st 2015 via Hootsuite

GG: Shares tool website at Broad: http://t.co/6ugvqlV8nU #AACR15

11:09am April 21st 2015 via Hootsuite

GG: Summary con't: A whole toolbox is needed; site-spec error models needed; validation req adv methods; many tools avail #AACR15

11:08am April 21st 2015 via Hootsuite

GG: MutationsValidator shows validation on TCGA dataset. Summarizing: 'mutation det is not a solved problem' #AACR15

11:07am April 21st 2015 via Hootsuite

GG: 2nd use cov. & alt reads in validation tumor data, min cut-off, and estimate noise at the site. Then given cutoff, calc power #AACR1

11:06am April 21st 2015 via Hootsuite

GG: But can falsely invalidate (i.e. insuff coverage); also noisy sites may give false positives. So first deal with germline at 30x #AACR15

11:04am April 21st 2015 via Hootsuite

GG: A typical validation of mutation is: targeted seq of mutated cases; or all samples; or match w/RNA-Seq data #AACR15

11:03am April 21st 2015 via Hootsuite

GG: Showed recovery of ~40% more driver mutation recovery in CLL via the deTiN tool (TiN=Tumor in Normal) #AACR15

11:01am April 21st 2015 via Hootsuite

GG: deTiN method uses potential somatic mut and allelic imbalance of germline heterozygous sites. Showed validation data. #AACR15

11:01am April 21st 2015 via Hootsuite

GG: FFPE C>T from colon ca exomes. CAC>CCC, GAG>GGG 'from only one seq center'; developed 'Panel of Normals' (PoNFilter) #AACR15

10:55am April 21st 2015 via Hootsuite

GG: Results in '13 NAR ref http://t.co/2iLonNADBM Another one: A>T in CLL exomes Shows IGV of somatic mut; strand bias, refined tool #AAC

10:52am April 21st 2015 via Hootsuite

GG: Reviews effect of oxidized G, G>T mutations that contaminate prior datasets (found both in normal & tumor) from ultrasonication #

10:50am April 21st 2015 via Hootsuite

Gad Getz (Broad Inst MA) "Mutation detection and validation using massively parallel sequencing data" GG #AACR15

10:49am April 21st 2015 via Hootsuite

New post: Studying rare mutations with QuantStudio 3D | Behind the Bench Blog http://t.co/nmDfRvZKRZ #AACR15

9:40am April 21st 2015 via Hootsuite

AA: Reinforcing model - the protein lesion binds to DNA and can't get off; thus inhibitor is how resistance occurs. #AACR15

9:15am April 21st 2015 via Hootsuite

AA: Now onto CRISPR '14 Nature Biotech ref http://t.co/B6umidzFNh Has model of PARP1 w/GFP reporter, showing effect of knock-out #AACR15

9:14am April 21st 2015 via Hootsuite

AA: Data sugg's autoPARsylation, release from DNA; reduce PARP, reduce effects of inhibitor. 'PARP trapping is the causative lesion' #AACR15

9:10am April 21st 2015 via Hootsuite

AA: ESC's: ~10K genes xcr, about 10^5 cells needed. Mapping insertions: many into PARP1; these are null mutants (little activity) #AACR15

9:08am April 21st 2015 via Hootsuite

AA: Predating CRISPR, used haploid ESC/transposon screens. Embryonic stem cells - only one copy, easy to knock-out. ID pos'n by NGS #AACR15

9:06am April 21st 2015 via Hootsuite

AA: "Hallmarks of 'BRCA-ness" '04 Nature Rev Cancer ref http://t.co/0sbxc4JBZn mentioned #AACR15

9:04am April 21st 2015 via Hootsuite

AA:70K shRNA x 20K genes to look for determinants of sens to PARP - many genes ID'd in DNA repair pathways #AACR15

9:02am April 21st 2015 via Hootsuite

AA:Looking at resistance to PARP may give clues to combination therapy. Case1: recurrence showed restoration of BRCA2 ORF #AACR15

8:58am April 21st 2015 via Hootsuite

AA: Mechanism - restoration of BRCA2 ORF in PARPi resist cell lines '08 Nature ref http://t.co/OZ2r7BPM6A New mech of resistance #AACR15

8:56am April 21st 2015 via Hootsuite

AA:Collapsed replication fork not being repaired - synthetic lethality '09 NEJM ref http://t.co/pIFCFi7S1P #AACR15

8:55am April 21st 2015 via Hootsuite

AA:Synthetic lethality: cells w/ BRCA1/2, used Poly(ADP-Ribose) Polym. (PARP), involved with excision repair, single-strand breaks #AACR15

8:52am April 21st 2015 via Hootsuite

AA: Starts with BRCA2 dysfunction. 'What do you target when the target is absent?' '12 Natre ref http://t.co/jbmFqRnGSm #AACR15

8:50am April 21st 2015 via Hootsuite

Alan Ashworth (UCSF CA) "Drug resistance: PARP inhibitors” AA #AACR15

8:48am April 21st 2015 via Hootsuite

BV: Checkpoint inhibitors: Landmark '12 NEJM http://t.co/4egwOqfwBZ One CR cancer pt had complete & durable response 'a #'s game' #AACR1

8:41am April 21st 2015 via Hootsuite

BV: "Early detection may not have to be >very< early to make a difference"; the need to treat cancers when they are small. #AACR15

8:39am April 21st 2015 via Hootsuite

BV: 'As you know, we are great at curing mice': new drug has to cure >100 mice with any chance to help in humans w/advanced disease #AACR

8:37am April 21st 2015 via Hootsuite

BV: The problem is sequential administration of chemotherapy; advocates combinations 'from the get-go' #AACR15

8:34am April 21st 2015 via Hootsuite

BV: Not a result of general resist. of stem cells, or genetic instab. in tumors. Impt clinical implications: use combination agents #AACR15

8:32am April 21st 2015 via Hootsuite

BV: 100's to 100K resistant cells prior to treatment. Movie of cell expansion over time: Billions of metastatic cells, then 2nd rnd #AACR15

8:28am April 21st 2015 via Hootsuite

BV: (Fifth time I'm seeing the Wagel photo of non-durable response) #AACR15

8:27am April 21st 2015 via Hootsuite