Tibbitts: Myrna: uses Bowtie + R for interval calcs; Hadoop running in the cloud "a little bit of work.. works well" #RNASeq2014
12:00pm June 19th 2014 via Hootsuite
Tibbitts: Universal sample ID (carried throughout); number of vendors for seq; cloud pipelines (Omicsoft) still TBD; Myrna #RNASeq2014
11:55am June 19th 2014 via Hootsuite
Tibbitts: Plan to do long-term archiving by pushing FASTQ files into 'cold storage'; $100/TB/year for this #RNASeq2014
11:51am June 19th 2014 via Hootsuite
Tibbitts: Decided to outsource seq; keep data in the cloud; use MSFT & AMZN; hard drives shipped from seq vendor to them #RNASeq2014
Tibbitts: Legacy SQL systems could not keep up; looked at Myrna/Hadoop for RNA-Seq data, and CouchDB to replace Oracle #RNASeq2014
11:48am June 19th 2014 via Hootsuite
Thomas Tibbitts (Infinity Pharmaceuticals): "Exploring scalable options for processing and storing of RNA-Seq data" #RNASeq2014
11:44am June 19th 2014 via Hootsuite
Lowe:Q&A: What about sizing of lib's for ENCODE? Known but this is a tool for screening and discovery #RNASeq2014
Lowe:Q&A: BAM uploads okay, but they prefer their own alignment to insure quality; integrated mRNA, miRNA, DNA var, ChIP, Me-Seq #RNASeq
11:42am June 19th 2014 via Hootsuite
Lowe: Data submitted for publ; able to now sequence through base mods with new method, showing cp to TCGA #RNASeq2014
11:37am June 19th 2014 via Hootsuite
Lowe: Showed some other examples of novel tRNA expression data from public sources; Demethylation library for mod RNA detection #RNASeq2014
11:35am June 19th 2014 via Hootsuite
Lowe: Hurto 2011 reference: http://t.co/tMombxxDpr Other roles for tRNAs #RNASeq2014
11:32am June 19th 2014 via Hootsuite
Lowe: Now looking at the same tRNA in ENCODE ncRNAs across cell lines: a pattern of 3' fragment in 2 of them; alt role of tRNAs #RNASeq2014
11:31am June 19th 2014 via Hootsuite
Lowe: Matching analysis to the library prep kit; not re-inventing 'yet another pipeline'. #RNASeq2014
11:30am June 19th 2014 via Hootsuite
Lowe: Maverix: 'Analytic kits' that anyone can use; may not be perfectly optimized kit, but 90% there w/o the opp'y cost #RNASeq2014
11:29am June 19th 2014 via Hootsuite
Lowe: Points out more fragments are there in Prostate and Br cancer tissues of these tRNAs in the TCGA datasets #RNASeq2014
11:25am June 19th 2014 via Hootsuite
Lowe: tRNA fragments relevant to cancer. 2013 PNAS: http://t.co/ZNCL3OOibG But only in cell lines; cp to TCGA? @MaverixBiomics #RNASeq2014
11:23am June 19th 2014 via Hootsuite
Lowe: Referred to a figure from this 2012 Turchinovich review http://t.co/8foHjQcmMb #RNASeq2014
11:21am June 19th 2014 via Hootsuite
Todd Lowe (Maverix): "Beyond known microRNAs: exploring the rest of the small RNA transcriptome" #RNASeq2014
11:19am June 19th 2014 via Hootsuite
Haynes:Q&A: Fusion param's take into account freq of 3'/5' partners #RNASeq2014
10:06am June 19th 2014 via Hootsuite
Haynes: A:(con't): 105-110bp library insert size. #RNASeq2014
10:05am June 19th 2014 via Hootsuite
Haynes:A (con't): all the 'junk on the right' were still useful (as low as ~20% exonic, ~10% intronic, ~75% intergen) #RNASeq2014
10:04am June 19th 2014 via Hootsuite
Haynes:Q&A: Can intragenic sequence be gDNA contam? Bar chart of exonic/intronic/intergenic, all libraries were used #RNASeq2014
10:03am June 19th 2014 via Hootsuite
Haynes: S Crosby comment before a question: "Fantastic talk". I agree. #RNASeq2014
10:01am June 19th 2014 via Hootsuite
Haynes: Training / test cohorts: models of FFPE trained very good sens. (.95) and spec (0.61) to FNA test set #RNASeq2014
9:59am June 19th 2014 via Hootsuite
Haynes: PC1/2 nice division between benign / malignant; also FNA vs FFPE (due to blood contamin of FNA) #RNASeq2014
9:58am June 19th 2014 via Hootsuite
Haynes: Applied to 123 FFPE including orig 68; also 65 FNAs retrospective. Nice concordance bet. targeted to WT RNA-Seq #RNASeq2014
9:57am June 19th 2014 via Hootsuite
Haynes: 41 gene signature migrating to targeted RNA-Seq assay, using targeted multiplex PCR approach. Only 50K reads/sample #RNASeq2014
9:55am June 19th 2014 via Hootsuite
Haynes: Of 47% 'no mutation' of malignants; 50% novel fusions, 33% no fusions, 17% published. Manuscript in prep #RNASeq2014
9:54am June 19th 2014 via Hootsuite
Haynes: "15 fusions spec to malignancies; remaining (of the 26) are likely trans-splicing" (trace levels) #RNASeq2014
9:53am June 19th 2014 via Hootsuite
Haynes: Sees concordance to TCGA cohort; ID 40K fusions in 68 samples (lots of FPs); 41 passed filtering; 26 conf. by Sanger #RNASeq2014
9:52am June 19th 2014 via Hootsuite
Haynes: 68 FFPE blocks of resected tumors (not FNA), about half malignant; about half of those had unknown drivers #RNASeq2014
9:49am June 19th 2014 via Hootsuite
Haynes: Thyroid cancer FNA samples: many are indeterminate via cytometry and single-gene approaches #RNASeq2014
9:47am June 19th 2014 via Hootsuite
Haynes: From 500 samples, success rate downstream on the order of 90% #RNASeq2014
9:46am June 19th 2014 via Hootsuite
Haynes: 50M reads 2x50 PE's. Their report is called 'Surasight'. They use chimerascan 2011 pub: http://t.co/bdfEZDgDK9 #RNASeq2014
Haynes: Algorithm choice for fusion breakpoints in FFPE matters; they have developed 3'/5' sensitive params. #RNASeq2014
9:44am June 19th 2014 via Hootsuite
Haynes: Gene fusions from FFPE: high FP (homologous genes); trans-splicing in normal cells occur; passenger fusions; short fx #RNASeq2014
9:43am June 19th 2014 via Hootsuite
Haynes: Modeling as nascent isoform cp to not-modeled: can change results. Data suggests unprocessed RNA picked up #RNASeq2014
9:42am June 19th 2014 via Hootsuite
Haynes: (QFI is a TM of Asuragen). Elevated intronic coverage from FFPE; most pipelines ignore it; ILM pulls out exonic #RNASeq2014
9:40am June 19th 2014 via Hootsuite
Haynes: "FFPE of low quality can be rescued" - by adding more. High quality: more not helpful (100ng to 600ng) #RNASeq2014
9:39am June 19th 2014 via Hootsuite
Haynes: QFI is qPCR-based, best predictor of FFPE whole transcriptome data quality. 100ng to 600ng chart shown... #RNASeq2014
Haynes: FFPE challenges on QC: %DV200 on BioAnalyzer by 'unique exonic r2': 0.17, 0.31, 0.34, 0.66 (BA, nanodrop, Qubit, QFI) #RNASeq2014
9:38am June 19th 2014 via Hootsuite
Haynes: FFPE can look 'good' but function poorly; also can look poor but function well. #RNASeq2014
9:36am June 19th 2014 via Hootsuite
Haynes: FFPE quality varies on 2 axes: irrev. chemical modifications & extent of degradation #RNASeq2014
Brian Haynes (Asuragen): "RNA-Seq of FFPE specimens for biomarker discovery & development" #RNASeq2014
9:33am June 19th 2014 via Hootsuite
Li:Q&A: The RNA amplification method was NuGEN's Ovation. #RNASeq2014
9:31am June 19th 2014 via Hootsuite
Li: Points out need to narrow down the inherent sample hetergeneity; mentions existing compute infrastructure & cloud movement #RNASeq20
Li: Doesn't recommend RNA ampl (? what method?); strandedness consistency important metric; some SW not sens. to strand #RNASeq2014
9:28am June 19th 2014 via Hootsuite
Li: lincRNA example cp protocols: varying results dep on library prep; miRNA picked up in all methods #RNASeq2014
9:25am June 19th 2014 via Hootsuite
Li: mRNA reads limited to 'small region': picked up via RT-PCR but not RNA-Seq; mRNA could be tied up in lipid. #RNASeq2014
9:23am June 19th 2014 via Hootsuite
Li: Case 3: Looking at RNA in exosome compartment; cp. protocols, +/- ampl.; different frag; 3 seq platforms (SOLiD, Ion + ILM) #RNASeq2014
9:22am June 19th 2014 via Hootsuite
