Erlich: He took 10 ng and did 25 cycles PCR and 10 rounds. Showed a movie thus encoded and decoded, earliest movie made (a train) #AGBT18
2:30pm February 14th 2018 via Hootsuite
Erlich: The problem with DNA retrieval - you use it up. Shows his 3y old: wants an unlimited number of songs. PCR brings errors #AGBT18
2:28pm February 14th 2018 via Hootsuite
Erlich: Their hints are 'Luby Transform'; he shows a movie shown here: https://t.co/9IHOOAecdn #AGBT18
2:27pm February 14th 2018 via Hootsuite
Erlich: Screening - check GC content, homopol, generate FASTA for synthesis. For 8 hints, only 1 used. #AGBT18
2:24pm February 14th 2018 via Hootsuite
Erlich: Strategy called DNA fountain. Hints about a few random digits, like a Sudoku puzzle. '17 ref https://t.co/LVBy7N2f6v #AGBT18
2:23pm February 14th 2018 via Hootsuite
Erlich: Homopolymers or high GC makes some DNA better than others. Challenge of retrieving: only 0.83 bits per nucleotide (nt) #AGBT18
2:21pm February 14th 2018 via Hootsuite
Erlich: #AGBT18 Church's '12 Science paper on DNA storage https://t.co/Iaud5pXYWQ Encoding isn't trivial...
2:20pm February 14th 2018 via Hootsuite
Erlich: DNA won't go obsolete, extremely dense, and highly robust. On the last point: can read ancient DNA #AGBT18
2:18pm February 14th 2018 via Hootsuite
Erlich: This paper did the math: how much DNA in biopshere? https://t.co/3EsSRGKkCH 12 orders of magnatude more than what we have #AGBT18
2:17pm February 14th 2018 via Hootsuite
Erlich: DNA is the 'ultimate storage device' - from virus through human. How much storage? #AGBT18
2:16pm February 14th 2018 via Hootsuite
Erlich: Price/Gb - leveling off in 2010: due to difficulties with the physics of storage. Also digital obsolescence. Ex: 8mm film #AGBT18
2:14pm February 14th 2018 via Hootsuite
Erlich: Also company called Family Tree - 86M records; bioRxiv preprint: https://t.co/QVGXQ0MqnI #AGBT18
2:13pm February 14th 2018 via Hootsuite
Yaniv Erlich (Columbia Univ NY, also CEO of My Heritage DNA) DNA Media #AGBT18
2:11pm February 14th 2018 via Hootsuite
Levy: #agbt18 math on samples of Twist WES on the NovaSeq https://t.co/Ww6IfsgAe2
1:11pm February 14th 2018 via Hootsuite
Levy: Oversimplifying indexing a bad idea - the indexing hopping problem. Dual indexing - the ideal set, no re-use. 3-6% rates #AGBT18
1:10pm February 14th 2018 via Hootsuite
Levy: 144 samples 4Gb, 45x coverage on S2; 192 samples, S4 10Gb. About $95 sequence cost on S2... #AGBT18
1:09pm February 14th 2018 via Hootsuite
Levy: Future: 96-well plate to 12 hybs, to 2 pools (6 hybs/pool), four pools per S4 flowcell; 10Gb/sample, 100x cov. He did the math #AGBT18
1:08pm February 14th 2018 via Hootsuite
Levy: All typicall 99.5% to 99.8%; not higher due to intronic variants. #AGBT18
1:07pm February 14th 2018 via Hootsuite
Levy: Metric is 100x coverage at each sampling rate (most sensitive measurement across the range). At full 500x: concordance >99% #AGBT18
1:06pm February 14th 2018 via Hootsuite
Levy: Efficiency was 'strong, a little better with HMW gDNA; Did downsampling to 4, 6, 9 and 15GB datasets. A way to show complexity #AGBT18
1:05pm February 14th 2018 via Hootsuite
Levy: Showed stats - a 500x WES (!), 500M reads/sample. 20% duplicate rate on 10GB; here in the low 20% range still. #AGBT18
1:02pm February 14th 2018 via Hootsuite
Levy: Did see small bias w/FFPE cp to HMW gDNA in indices, but 'still amazingly balanced' 10% for FFPE vs about 13-14% for HMW gDNA #AGBT18
1:00pm February 14th 2018 via Hootsuite
Levy: NovaSeq: 1 lane of S2; XP workflow; 93% bases at >Q30. 1TB run (more than Illumina states). Shows indexing representation #AGBT18
12:59pm February 14th 2018 via Hootsuite
Levy: Had 8 samples x 2; two HapMap samples, 6 germline samples, 4 FFPEs. All run as pools of 8; FFPE pooled with HMW gDNA #AGBT18
12:58pm February 14th 2018 via Hootsuite
Levy: His criteria - workflow, input: 50ng input, two purifications. frag/endrepair, add-A; ligate, purify; PCR, purify, QC. #AGBT18
12:57pm February 14th 2018 via Hootsuite
Levy: Shows the diversity of 9 mos of thousands of samples. Wide range of age and quality of FFPE samples. #AGBT18
12:56pm February 14th 2018 via Hootsuite
Levy: Oversees both clinical and research labs; 'that huge number is roughly the number of times @mason_lab (Chris… https://t.co/TsbjWFqZQt
12:55pm February 14th 2018 via Hootsuite
Shawn Levy (HudsonAlpha AL) Deep exome sequencing in FFPE samples #AGBT18
12:53pm February 14th 2018 via Hootsuite
Twist: 95% sens for NA12878; WGS requires 15x (56GB); Twist WES is 16.5x (1.3GB). Off target 15%; 3% duplication rate at 50x #AGBT18
12:46pm February 14th 2018 via Hootsuite
Twist: '14 ref https://t.co/yOBY6Kbdpx shows equivalent coverage metrics for same sensitivity. #AGBT18
12:44pm February 14th 2018 via Hootsuite
@h2so4hurts Emily didn't name the comparison here at #AGBT18. And I didn't get the current speaker's name, he's not… https://t.co/YSu6TIK8fY
12:43pm February 14th 2018 via Hootsuite in reply to h2so4hurts
LeProust: NGS QC of probes - shows what passes vs what doesn't. "You are invited to 'break up with your exome' on Valentine's day" #AGBT18
12:38pm February 14th 2018 via Hootsuite
LeProust: coverage as function of GC% looks very favorable. #AGBT18
12:30pm February 14th 2018 via Hootsuite
LeProust: dsDNA probe synthesis: melting and re-annealing increases accuracy. WES: 33MB targett. >99% target at 20x. cp to 93% other #AGB
12:29pm February 14th 2018 via Hootsuite
LeProust: Biased oligo pools - starts with synthesis. Amplification of oligos - larger bias. Their amp method - more uniform #AGBT18
12:28pm February 14th 2018 via Hootsuite
RT @TorontoGenomics: @coregenomics Courtesy @coregenomics
12:26pm February 14th 2018 via Hootsuite
LeProust: Compares WGS to TE (target enrichment); $7.4K storage cost/year for WGS shown. #AGBT18
LeProust: 'Everyone hates cloning.' 9c/base for gene synthesis. 15K genes/mo shipped. Ginko deal: 1B bases to buy. #AGBT18
12:24pm February 14th 2018 via Hootsuite
LeProust: Points out 96 oligos to 1 gene; theirs >1M oligos to 9,600 genes. Genes, oligo pools, libraries, NGS #AGBT18
12:23pm February 14th 2018 via Hootsuite
Emily LeProust (CEO Twist Bioscience): Has raised $450M, offices in SF, SSF, SD, Tel Aviv. Introduces the big vision of new apps #AGBT18
12:22pm February 14th 2018 via Hootsuite
Martignetti: ...for endometrial cancer. You know there are driver mutations... #AGBT18
12:06pm February 14th 2018 via Hootsuite
Martignetti: 1st evidence of ability of an NGS-based approach coupled with targeted sampling can prospectivley detect early-stage #AGBT18
12:05pm February 14th 2018 via Hootsuite
Martignetti: 10mos after: postmenopausal bleeding; complex hyperplasia, lots of surgery, shows the slide w/a mark - the cancer #AGBT18
12:01pm February 14th 2018 via Hootsuite
Martignetti: One pt - found five driver muts in four genes; AFs 5% down to 1.3%. No evidence, molecular features - what to do? #AGBT18
12:00pm February 14th 2018 via Hootsuite
Martignetti: 51/102 women also had driver or potential driver mutations. AFs were low - 1-30% range, ave 3.0% #AGBT18
11:58am February 14th 2018 via Hootsuite
Martignetti: Slides from two samples - inside polyps (very small). AF's range dwon to 1.7%, 2.5% in the lavage. Similar to cell AF #AGBT18
11:57am February 14th 2018 via Hootsuite
Martignetti: Using a Swift Biosci cfDNA panel: 12 genes. Shows 7 women (10%), six were stage 1A, six grade 1 (earliest). #AGBT18
11:56am February 14th 2018 via Hootsuite
Martignetti: This work published in '16 https://t.co/lgeQF4gOU0 107pts hysteroscopy, collection of fluid prospectively #AGBT18
11:55am February 14th 2018 via Hootsuite
Martignetti: Can it be used for endometrial early detection? But limited amount, what tissue? How about a lavage... #AGBT18
11:53am February 14th 2018 via Hootsuite
Martignetti: There are no molecular tests. '13 TCGA project: https://t.co/4liaeCworn Can liquid biopsy be used? #AGBT18
11:52am February 14th 2018 via Hootsuite
