Data Science newsletter – November 10, 2021

Three years ago, neurologist Tim Yu stunned colleagues at the American Society of Human Genetics (ASHG) annual meeting with news that in just 10 months, his team at Boston Children’s Hospital had developed, tested, and administered a drug tailored to the unique genetic mutation causing a 7-year-old girl’s deadly brain disease. The story of milasen (named for the girl, Mila) raised hopes for more such personalized treatments made from short strands of RNA, known as antisense oligonucleotides (ASOs), that can overcome a gene’s flaws.

At this year’s ASHG meeting, held online last month, Yu shared a sobering update: Mila died earlier this year—tragic evidence that for neurodegenerative diseases, “it’s critical to intervene early,” he told a session.

But Yu’s group and others are now extending the customized ASO approach, known as “n of 1” because the drug is designed for just one or a few patients.

Giving social media giants more ways to curate communication or demanding that they whack more moles won’t solve the problem of perverse incentives. Instead, we should seek to foster business models that generate profits by serving people, not advertisers.

You’ve made predictions recently about budding revolutions in Internet architecture with Web 3 and the future of creator-centric economies with NFTs which have made me wonder about something I had asked Ben Horowitz in a recent interview: There are exciting things happening in the world of bits and computing without a doubt, but is there anything left to be excited about in the world of atoms? Can things like crypto help us live like The Jetsons?

Great question. A couple of points.

First I’d argue that the digital world and the physical world are already deeply interconnected and getting more so everyday. The internet is a global brain that increasingly orchestrates the global body. One of the myths around Jetsons-like robots is that automation happens visibly, with a 1:1 correspondence between the thing replaced and the thing replacing it. In fact, automation arrives in far subtler, more mundane ways: a database installation here, an API integration there. The internet is a network of networks, some public, like the web or Twitter, and some private, within social groups or companies. Collectively these networks orchestrate more and more of what happens in the physical world.

It’s no longer about whether we’ll transition to 100% renewable generation, but rather how to do it quickly and keep everyone’s lights on in the meantime. Energy storage is increasingly the answer to this question, and as such, a cornerstone in the roadmap to a cleaner, distributed energy infrastructure. Storage helps operators prepare for and capture more value from the increasing penetration of intermittent resources, while efficiently supporting system reliability.

Incremental steps with storage matter. Just integrating storage alone with internal combustion engines offers considerable potential for fuel and cost savings. Power producers can save up to 6% in annual fuel savings, while also reducing emissions just by adding energy storage. This facilitates a more seamless setup for adding renewable assets to a generation mix.

But pairing energy storage with software opens the door to both net zero and grid stability even further.

Electro-physiological sensing devices are becoming increasingly common in diverse applications. However, designing such sensors in compact form factors and for high-quality signal acquisition is a challenging task even for experts, is typically done using heuristics, and requires extensive training. Our work proposes a computational approach for designing multi-modal electro-physiological sensors. By employing an optimization-based approach alongside an integrated predictive model for multiple modalities, compact sensors can be created which offer an optimal trade-off between high signal quality and small device size. The task is assisted by a graphical tool that allows to easily specify design preferences and to visually analyze the generated designs in real-time, enabling designer-in-the-loop optimization. Experimental results show high quantitative agreement between the prediction of the optimizer and experimentally collected physiological data. They demonstrate that generated designs can achieve an optimal balance between the size of the sensor and its signal acquisition capability, outperforming expert generated solutions. [full text]

The goal is a full readout of all the proteins being made in a cell. No technology can do that yet, but mass spectrometry comes closest. And as the sensitivity of mass spectrometers has improved, researchers have realized that they can finally achieve single-cell proteomics.

Nikolai Slavov, a systems biologist at Northeastern University, is a pioneer in single-cell proteomics. As a graduate student and postdoc, he analyzed DNA and RNA in single cells. He wanted to do similar measurements of proteins, but “it went against the grain to think that mass spectrometry would have enough sensitivity to analyze single cells at high scale,” Slavov says.

The new Institute for Infectious and Zoonotic Diseases, launched by the School of Veterinary Medicine, leans on Penn’s strengths in immunology and infectious disease to prepare for emerging threats to animal and human health.

Research by scientists from UCLA and Lawrence Livermore National Laboratory strengthens the case that climate change has been the main cause of the growing amount of land in the western U.S. that has been destroyed by large wildfires over the past two decades.

Rong Fu, a UCLA professor of atmospheric and oceanic sciences and the study’s corresponding author, said the trend is likely to worsen in the years ahead. “I am afraid that the record fire seasons in recent years are only the beginning of what will come, due to climate change, and our society is not prepared for the rapid increase of weather contributing to wildfires in the American West.”

When Jessica Toh had her first child, who proved to be an erratic sleeper, she wanted to use her background in computer science and statistics to understand why.

In 2017, she started Huckleberry, an app that aims to leverage AI to predict babies’ sleep schedules and help caretakers get some rest.

The Irvine-based app announced it raised $12.5 million on Wednesday led by Morningside Ventures, a science-focused venture capital firm.

Google has released a cloud-based “sandbox” where public agencies, along with laboratories and university IT departments, can test and learn new technologies.

The new tool, called RAD Lab, makes its debut as more governments turn to cloud tools and software as a service — a big factor in the growing amount of investment and merger and acquisition activity in the government technology sector.

“For state and local governments, the pandemic has driven an upsurge in innovation, and that shift is here to stay,” said Emma Fish, head of education programs and business development of Cloud Public Sector at Google. “Many (of those agencies) are thinking about how to upscale and support technology teams.”

DLSS actually stands for deep learning super sampling, and it’s a type of video rendering that renders frames at a lower resolution to boost your FPS, and then gets deep learning AI to upscale the frames so everything looks nice. I just… don’t really understand how that actually works.

But I know what it allows me to do: DLSS in Back 4 Blood lets me play the game in 4K and still get a high framerate, so performance isn’t impacted even when scores of zombies are running around and bullets, explosives and gore are filling the air.

In short, DLSS is taking on the hard work so that I can get a smooth framerate without having to turn down all of my graphics settings.

NASA has announced a $177 million Earth science mission led by Colorado State University that will study the behavior of storms in the Tropics, with the goal of better representing these storms in weather and climate models.

The mission will be a collection of three small satellites, flying in tight coordination, and is called Investigation of Convective Updrafts (INCUS). It is expected to launch in 2027 as part of NASA’s Earth Venture Program.

INCUS’ principal investigator is Susan van den Heever, professor in the Department of Atmospheric Science, whose expertise is in cloud physics, cloud dynamics, and mesoscale meteorology and modeling.

A new generation of networked, cooperating robots being developed by researchers in the Center for Connected Autonomy and Artificial Intelligence (ca-ai.fau.edu) at Florida Atlantic University’s College of Engineering and Computer Science and Institute for Sensing and Embedded Network Systems Engineering (I-SENSE) is all about “teamwork.” The Center received a $1 million grant from the National Science Foundation (NSF) to develop the nation’s first-of-its-kind testbed platform that connects robots using extremely high-speed millimeter wave links. These high-frequency radio waves (30 to 300 gigahertz), until more recently, were primarily used for remote sensing, radio astronomy, automotive radar and security screening.

With a three-year grant from the Department of Energy, a team of researchers from Johns Hopkins and partner organizations are launching the Synthesis and Processing Informed by Rational Algorithmic Learning Center, or SPIRAL, a multi-institution effort to harness the power of artificial intelligence to accelerate the optimization of low-cost materials for solar energy.

Johns Hopkins faculty members Paulette Clancy, head of the Whiting School of Engineering’s Department of Chemical and Biomolecular Engineering, and Rigoberto Hernandez, professor of chemistry in the Krieger School of Arts and Sciences, will help lead the project’s machine learning component.

In the largest single donation in its history, the nonprofit Santa Fe Institute will receive $50 million from legendary investor Bill Miller. The gift will advance the Institute’s pioneering science of complex systems by growing its research community and expanding the facilities in which it works.

Complexity science seeks to find the organizing patterns at the heart of systems with a multitude of adaptive parts — from economies to ecosystems. Many scientists, including the late physicist Stephen Hawking, predicted that the 21st century would be “the century of complexity,” when science would build on the foundational laws of physics by understanding “how the laws fit together, and what happens in extreme circumstances,” Hawking said.

“This gift comes at a moment when the world needs radically new ideas and quantitative frameworks to engage with the growing connectivity and complexity of life and the accelerating pace of change in both technology and society,” says SFI President David Krakauer.