Big Data A Discovery Dinner College of Natural Sciences October - - PowerPoint PPT Presentation

Big Data

A Discovery Dinner College of Natural Sciences October 13, 2015

Presenters: Karl Gebhardt, Astronomy Lauren Ancel Meyers, Integrative Biology Stefano Tiziani, Nutritional Sciences Moderated by Hans Hofmann, Integrative Biology Mike Daniels, Statistics and Data Sciences

Hobby-Eberly Telescope Dark Energy Experiment The upgraded HET

(Karl Gebhardt, Project Scientist)

HETDEX is:

→ Blind spectrographic survey on 10m Hobby-Eberly Telescope → About 1 million galaxies around 10 billion years ago → About 2 million galaxies within 4 billion years ago → New instrument VIRUS with 33,500 fibers over focal plane → Generate about 0.3 Pb/yr for 5-10 years → Goal is to measure the expansion rate of Universe.

Data Issues:

→ Transfer, real-time analysis, storage on TACC → Access to a broad community → Develop techniques for understood processes and new exploratory studies

ESSENCE of HETDEX

380,000 years after the Big Bang Today: 13.7 billion years after the Big Bang

• There is a characteristic scale

imprinted in the distribution of matter.

• It acts like a fingerprint that we

use to trace over cosmic time.

• Requires about one million

galaxies in order to see it.

HETDEX Analysis Issues

• How to handle many Pb of data. We are working with TACC
• n this. All data will be public.
• Base programs are all in c++, with python wrappers for

metadata and access. Need developers who understand statistics, astrophysics, coding.

• As is always the issue, large-scale systematics will be the

limiting factor for the uncertainties.

• Example case: We will have 1e6 objects detected. We will

have 1e9 spectra of blank sky. Correlate detections against blank sky to pull out spatial signal from fainter sources.

• Example case: Detection of cosmic web, which is

correlation over full region (3 Pb) of unknown pattern.

• Example case: Real-time analysis to look for supernovae,

requires fast transfer, analysis, and feedback.

Big data and the race to stop global pandemics

Lauren Ancel Meyers Department of Integrative Biology Department of Statistics & Data Sciences The University of Texas at Austin

Where are outbreaks spreading today?

Where will they be tomorrow?

How can we stop them?

 To obtain a metabolic signature of clinical samples

• Early detection
• Disease stratification
• Disease progression
• Therapy outcome

The Tiziani Research Group

 To identify novel top-hit most effective compound combinations for targeting the tumor microenvironment  To better understand cellular metabolism and its link to human diseases

The Tiziani Research Group

This urine wheel was published in 1506 by Ulrich Pinder, in his book Epiphanie Medicorum used to diagnose disease

The Metabolome

Thousands of features Metabolites/lipids validated Not detectable yet… Metabolites/lipids potentially identified 103 104 105 106 102

Targeting the Metabolome is Challenging

Stable Isotope Assisted-Metabolomics

13C 1H

• NMR- and MS-based metabolic

studies

(intra- and extra-cellular)

• Metabolic fluxes analysis
• Constraining simulations
• 13C, 15N labeled nutrient
• High-Throughput Screening

(e.g. ATP and ROS assays 384 well plate)

• High-Content NMR- and MS-based

metabolic screening

(96 well plate, flasks)