20170402 Crop Innovation and Business - Amsterdam
- 1. Allen Day, PhD // Science Advocate // @allenday // #genomics #ml #datascience
- 2. GOOGLE CONFIDENTIAL Google Cloud Run your apps on the same system as Google
- 4. Generate Marker Fingerprint Select & Recombine Sample tissue Breeding Genotyping Lab Extract DNAAnalyze & Model Data Grow Marker-Assisted Breeding Rapidly Increases Frequency of Favorable Genes Cloud ML TensorFlow
- 5. AI & ML what you need to know Machine Learning: Make Machines Learn Artificial Intelligence: Make Intelligent Machines programming a computer to be intelligent is hard programming a computer to learn to be intelligent is easier and progress is measurable
- 6. * Human Performance based on analysis done by Andrej Karpathy. More details here. Image understanding is (getting) better than human level ImageNet Challenge: Given an image, predict one of 1000+ of classes %errors
- 7. Deep Neural Networks: Algorithms that Learn ● Modernization of artificial neural networks ● Made of of simple mathematical units, organized in layers, that together can compute some (arbitrary) function ● more layers = deeper = more general ● Learn from raw, heterogeneous data
- 8. “Given an image, predict one of 1000+ of classes” Image credit: 360phot0.blogspot.com ImageNet Challenge
- 9. Released in Nov. 2015 #1 repository for “machine learning” category on GitHub TensorFlow
- 10. Style Transfer
- 11. Transfer Learning Quickly able to Learn New Concepts “t-rex”“quidditch” Learning like a Child: Fast Novel Visual Concept Learning from Sentence Descriptions of Images
- 12. TensorFlow powered Cucumber Sorter
- 13. Generate Marker Fingerprint Select & Recombine Sample tissue Breeding Genotyping Lab Extract DNAAnalyze & Model Data Grow Marker-Assisted Breeding Rapidly Increases Frequency of Favorable Genes Cloud ML TensorFlow
- 14. Genomics & Genetics Problems: How to Start Applying DNNs? Must-haves for deep learning: ● Lots of data: >50k examples, >1M examples ideal ● High-quality input and labels for training ● Label ~ F(data) unknown but certainly function exists ● High-quality prev. efforts so we know that DNNs are key ○ i.e. hard to solve with classical statistical approaches SNP and indel calling from NGS data
- 17. Creating a universal SNP and small indel variant caller with deep neural networks Ryan Poplin, Cory McLean, Dan Newburger, Jojo Dijamco, Nam Nguyen, Dion Loy, Sam Gross, Madeleine Cule, Peyton Greenside, Justin Zook, Marc Salit, Mark DePristo, Verily Life Sciences, October 2016
- 18. DNN (Inception V3) Predicts True Genotype from Pileup Images { 0.001, 0.994, 0.005 } { 0.001, 0.990, 0.009 } { 0.000, 0.001, 0.999 } { 0.600, 0.399, 0.001 } Output: Probability of diploid genotype states { HOM_REF, HET, HOM_VAR } Raw pixels Input: Millions of labeled pileup images from gold standard samples
- 19. DeepVariant #1 in PrecisionFDA Truth Challenge v2 => v3 truth set for unblinded sample Unblinded => blinded sample with v3 truth set 99.85 99.70 98.91
- 21. ⬇40% Data Center cooling energy ⬆15% Power Usage Effectiveness (PUE) Google’s Carbon-Neutral, Self-Optimizing Data Centers The Dalles, Oregon, USA
- 22. anezconsulting.com/precision-agronomy/ Agronometric Integration ● Satellite & UAV Images ● Geological Data ● Meteorological & Sensor Data ● Cultivar Data ● Other GIS Data ● Yield Data
- 24. Public Datasets Project https://cloud.google.com/bigquery/public-data/ A public dataset is any dataset that is stored in BigQuery and made available to the general public. This URL lists a special group of public datasets that Google BigQuery hosts for you to access and integrate into your applications. Google pays for the storage of these data sets and provides public access to the data via BigQuery. You pay only for the queries that you perform on the data (the first 1TB per month is free)
- 26. Marker-assisted selection for quantitative traits “Marker Assisted Selection” & “Quantitative Trait Locus” Occurrence in Literature is Increasing
- 27. GraphConnect SF 2015 / Graphs Are Feeding The World, Tim Williamson, Data Scientist, Monsanto https://www.youtube.com/watch?v=6KEvLURBenM
- 28. GraphConnect SF 2015 / Graphs Are Feeding The World, Tim Williamson, Data Scientist, Monsanto https://www.youtube.com/watch?v=6KEvLURBenM
- 29. PubSub Queue Sequencer Reads Genomics APIs, Docker Revise Models Models Cloud ML MAB Enhance Percolate Streaming Sequencer Reads for Real-time Model Updates BigQuery
- 30. Google confidential │ Do not distribute Google is good at handling massive volumes of data uploads per minute users search index query response time 300hrs 500M+ 100PB+ 0.25s
- 31. Google confidential │ Do not distribute Google can Handle Massive Amounts of Genomic Data uploads per minute users search index query response time 300hrs 500M+ 100PB+ 0.25s ~6 Maize WGS >100x US PhDs ~1M WGS 0.25s
- 32. PubSub Queue Genomics APIs, Docker Revise Models Models MAB Enhance Percolate Streaming Sequencer Reads for Real-time Model Updates Who Else Needs This? Sequencer Reads Cloud ML BigQuery
- 33. New Public Dataset: 1K Cannabis cloud.google.com/bigquery/public-data/1000-cannabis Blog Post @ Medium: DNA Sequencing of 1K Cannabis Strains publicly available in Google BigQuery Open Source: https://github.com/allenday/bfx-seq Revise Models DNA Reads
- 34. Build What’s Next Thank You! Allen Day, PhD // Science Advocate // @allenday // #genomics #ml #datascience