SLIDE 1 1
Deep Learning for Shallow Sequencing
Johnny Israeli Nvidia Genomics Group GTC 2018
SLIDE 2 2
Talk Outline
- A Study of Breast Cancer
- Sequencing Technologies
- Deep Learning for Genomics
- Enhancing Sequencing with Deep Learning
SLIDE 3
SLIDE 4 25 μm
100 Pa 2000 Pa
SLIDE 5 ATAC-seq profiling of genomic elements controlling gene expression
Map to genome
SLIDE 6
Control elements activated in stiff cells reveal the SP1 DNA motif
SLIDE 7
Knock down of SP1 in stiff cells removes invasive phenotype Soft Stiff SP1 Knock down
SLIDE 8 8
Sequencing Technologies
SLIDE 9 Genome Analyzer IIx HiSeq 2500 MinION SmidgeION
SLIDE 10
SLIDE 11
Sequencers as Counting Devices
SLIDE 12
Sequencers as Counting Devices
SLIDE 13
Sequencers as Counting Devices
SLIDE 14
Sequencers as Counting Devices
SLIDE 15
Diverse sequencing-based technologies
SLIDE 16
SLIDE 17
Growth of single cell sequencing
SLIDE 18 18
Deep Learning for Genomics
SLIDE 19
DL4Genomics takes off 2015
SLIDE 20
DeepSEA (2015): Predicting effects of variations in the non-coding genome with deep learning
SLIDE 21
DeepVariant (2016): Detection of single nucleotide genomic variations and insertions/deletions with deep learning
SLIDE 22
Chiron (2017): Translating MinION measurements into DNA Sequence with Deep Learning
SLIDE 23
1,700 Models!
SLIDE 24 24
Enhancing Sequencing with Deep Learning
SLIDE 25
Is this the same ATAC-seq signal?
SLIDE 26
Is this the same ATAC-seq signal?
SLIDE 27 50M DNA Reads 5M DNA Reads
140- 18-
Enhancing ATAC-seq with Deep Learning
SLIDE 28 50M DNA Reads 5M DNA Reads
140- 18-
50M DNA Reads 5M DNA Reads
53- 8-
Enhancing ATAC-seq with Deep Learning
SLIDE 29 50M DNA Reads 5M DNA Reads
140- 18-
50M DNA Reads 5M DNA Reads Linear Model Enhancement
53- 53- 8-
Enhancing ATAC-seq with Deep Learning
SLIDE 30 50M DNA Reads 5M DNA Reads Linear Model Enhancement Deep Learning Enhancement 50M DNA Reads 5M DNA Reads
140- 18- 53- 53- 53- 8-
Enhancing ATAC-seq with Deep Learning
SLIDE 31 50M DNA Reads 5M DNA Reads
140- 18-
50M DNA Reads 5M DNA Reads Linear Model Enhancement Deep Learning Enhancement
140- 140- 140- 17-
Enhancing ATAC-seq with Deep Learning
SLIDE 32 50M DNA Reads 5M DNA Reads
140- 18-
5M DNA Reads Signal Enhanced by DL ~ 20M Signal!
Enhancing ATAC-seq with Deep Learning
SLIDE 33 More Sequencing Data Better Deep Learning Models Cheaper Signal Generation Better Data Enhancement
Augmenting Sequencers with Deep Learning
SLIDE 34 34
Genomics @ Nvidia (we are hiring!)
Joint work with
- Nvidia AI Infrastructure
- Nvidia Research
- Nvidia Healthcare
Nikolai Yakovenko
Thank you for listening! Questions? jisraeli@nvidia.com