Data Science newsletter – March 7, 2021

Conservation plans increasingly include goals to maintain a connected network. For example, planners might design a linkage between two conserved areas, or the parties to the Convention on Biological Diversity might set targets for a well-connected system of protected areas for each nation. Here we describe 35 metrics that can quantify connectivity of focal patches or of networks and monitor changes over time in an ecoscape (landscape or seascape). The connectivity metrics fall into four categories: (1) structural connectivity metrics derived from binary maps and species-nonspecific spatial functions, (2) connectivity metrics derived from binary maps and species-specific population sizes and dispersal functions (3) metrics derived from multi-state maps and species responses to those states, and (4) metrics of functional connectivity reflecting observed flow of organisms or genes. We provide a decision tree to select which of these metrics are most appropriate for a given conservation goal and broad ecoscape context. Functional connectivity metrics may be preferred if conservation is focused on particular species or if data are available to parameterize models for a suite of species that represent needs of the focal biota. However, with climate change, ecoscapes need to facilitate movements of all species that need to adapt by shifting their ranges. Because an intact network of relatively natural areas may support movement for many species, structural metrics that consider the human footprint should be used in all coarse filter approximations of functional connectivity in shared ecoscapes.

Making cheese leaves a lot to chance as a batch could be ripened for months or even years before a problem is discovered, which could send a prized batch of cheddar to be sold off cheap as an ingredient for processed cheese.

It’s part of why cheese is so complex and expensive to make – a factory could invest lots of time and money into what they think will be a top-graded batch, only to discover it’s a flop when it’s too late to fix.

But new research from RMIT University allows quality to be checked much earlier and more precisely in the process, giving manufacturers a better chance to react to issues with the ripening process.

March 1 is National Peanut Butter Lovers Day, and fans of the lunchtime staple will certainly be celebrating. But what’s your preference when it comes to peanut butter? Whether or not you like it creamy or crunchy may actually say more about you than you think, according to a new survey. More importantly, it might also reveal if your relationship is headed for a sticky situation.

The OnePoll study examines the peanut butter and snacking preferences of 2,000 Americans – evenly split by their preference of crunchy vs. creamy. Researchers say that those who enjoy the crunchy kind have a cheerier outlook on life.

Sixty-three percent of those who prefer crunchy peanut butter describe themselves as optimists, compared to 56 percent of those who prefer creamy. Other personality traits for creamy fans include being more of an early bird and being more introverted. Their crunchy counterparts on the other hand are more likely to be night owls and extroverts.

When Teddy Goetz—a fourth-year medical student at Columbia University—applied to residency programs in October 2020, he felt as though he had no choice but to out himself as transgender. “I had to put my birth name all over my application because of my publications, and that was really upsetting,” he says. He changed his legal name to Teddy last year. But many of his papers listed him using his birth name.

Before submitting his applications, Goetz had contacted every journal he’d published in—14 in total—to request they change his name. Two journals offered to change his name and issue a correction notice. Many others didn’t have a policy to deal with author name changes and refused to change his name without one. It was disheartening, but he continued to press the journals to accommodate his request. Now, his name is changed or in the process of being changed on all but one of his publications. “It’s been a very long process and involves a lot of … labor, time, energy, attention, massive spreadsheets,” he says. But it’s worth it. “My legacy should not be the name that isn’t mine; the legacy should be mine.”

Goetz is part of an informal group of transgender scientists who have been pushing for changes to the scientific publishing industry to make it more inclusive—not only for trans scientists, but also for others who change their names midcareer, for instance because of a change in marital status or religion.

With so many questions being raised by the growing number of mutations, Yahoo News turned to resident medical expert Dr. Kavita Patel for answers. (The following interview was edited for clarity.)

Yahoo News: How worried should Americans be about coronavirus variants now circulating?

Dr. Kavita Patel: I think people should be worried. There’s a large number, a majority of the population that has not been vaccinated. They should be very worried, because they are prime targets for these viruses with the variants to reproduce.

Remember, the goal of a virus is not to kill people, it’s actually just to continue to stay alive, and the only way it does that is by infecting people. People who are not vaccinated should be incredibly worried, which is why I, in turn, am very worried about the variants as I watch now 12 states and counting, very big states including Texas and Florida, lifting any sort of mask requirements or leaving it to individuals or businesses. That’s a group that should be very worried.

Microsoft and Amazon have embarked on a frantic race for dominating the cloud machine learning (ML) ecosystem. Every year, both technology giants release dozens of new additions to their ML platforms. Last week, it was Microsoft’s turn to dominate the news. During its Ignite conference, Microsoft unveiled an overwhelming number of new ML services that expand the capabilities of its already impressive ML offering.

A Nature poll shows that a year of online research conferences has brought big benefits, but blending them with in-person meetings in future will be a challenge.

Could the office be detrimental to our productivity? Since many companies shifted to remote working arrangements, workday hours have shrunk, however employee productivity remained steady and even increased, according to data collected by Prodoscore, provider of employee visibility and productivity intelligence software. Looks like we’re accomplishing more in less time.

The study found that calendar time had dropped nearly 23% compared to the year prior, registering an average of 116 minutes of productive time per day. However, productivity levels from May to August 2020 were up 5% compared to the same timeframe in 2019.

Prodoscore also pinpointed when its users are getting more done, with the window of greatest productivity from 10 a.m. to 1 p.m. Tuesday is the most productive day of the week, followed by Wednesday and Thursday. And it’s no surprise that productivity is the lowest on Friday.

With a multidisciplinary approach cutting across systems biology, microbiology, evolutionary biology and other disciplines, researchers analyzed massive amounts of genome sequence data generated from more than 400 samples. They investigated the temporal and combinatorial patterns in which mutations accumulated in both organisms over 1,000 generations, mapped lineages through high resolution single cell sequencing, and characterized the fitness and cooperativity of pairings of their individual isolates.

The team uncovered striking evidence that mutations accumulated during evolution generate positive genetic interactions among rare individuals of a microbial community. These genetic interactions increase cooperativity within these rare microbial assemblages, enabling their persistence at very low frequency within a larger productive population. In addition, researchers discovered one of the first examples of parallel evolution, i.e., accumulation of mutations in similar genes across independently evolving populations, underlying the evolution of both organisms in a mutualistic community.

The land-grant university (LGU) system serves the nation by fostering excellence in research, delivering game-changing breakthroughs and training tomorrow’s agricultural, food, nutrition and bioeconomy innovators. Our enterprise accelerates technology adoption, growth of the agricultural and food marketplace, entrepreneurship and public-private partnerships.By doing so, we have supported American jobs, exports and economic growth –even as public sector funding for agricultural research lags behind that of our international competitors.

Still, the United States (U.S.) risks losing the ability to compete internationally if we ask our researchers and educators to conduct and deliver 21st century results in facilities from the 1950s and 1960s.

Fears of a dystopian urban world where people dodge heavy, fast-moving droids are colliding with the aims of robot developers large and small — including Amazon and FedEx — to deploy delivery fleets. … Driving the news: States like Pennsylvania, Virginia, Idaho, Florida and Wisconsin have passed what are considered to be liberal rules permitting robots to operate on sidewalks — prompting pushback from cities like Pittsburgh that fear mishaps.

Scientists have long sought a way to use those submerged cables to monitor seismicity. After all, more than 70 percent of the globe is covered by water, and it is extremely difficult and expensive to install, monitor, and run underwater seismometers to keep track of the earth’s movements beneath the seas. What would be ideal, researchers say, is to monitor seismicity by making use of the infrastructure already in place along the ocean floor.

Previous efforts to use optical fibers to study seismicity have relied on the addition of sophisticated scientific instruments and/or the use of so-called “dark fibers,” fiber optic cables that are not actively being used.

Now Zhongwen Zhan (PhD ’13), assistant professor of geophysics at Caltech, and his colleagues have come up with a way to analyze the light traveling through “lit” fibers—in other words, existing and functioning submarine cables—to detect earthquakes and ocean waves without the need for any additional equipment. They describe the new method in the February 26 issue of the journal Science.

Among the many mysteries in medical science, it is known that minority and low-income patients experience greater pain than other parts of the population. This is true regardless of the root cause of the pain and even when comparing patients with similar levels of disease severity.

As reported in Nature Medicine, researchers show that AI can more accurately and more fairly measure severe knee pain.

Today, when patients with knee pain visit the doctor, the severity of their osteoarthritis is rated on what is known as the Kellgren and Lawrence Grade (KLG). However, even for two patients with similar osteoarthritis and the same KLG score, low-income populations report more pain. Consequently, the underserved fail to qualify for knee-replacement surgeries and are more often treated with risky opioid painkillers.

The College of Human Ecology is working to address this knowledge gap with its Data Science and Programming Curriculum Initiative, an effort launched in 2019 to teach Human Ecology students how to use data and technology in their respective disciplines. To date, the initiative includes three classes that teach programming and data application in apparel design, molecular nutrition and public policy.

“Data science and programming are tools that are critical to many fields,” said Rachel Dunifon, the Rebecca Q. and James C. Morgan Dean of the College of Human Ecology. “My goal was to give students the chance to develop skills in data science and programming, and see how they can be brought to bear in our multidisciplinary, applied, problem-focused curriculum in Human Ecology.”

The chapter overviews the impact of COVID-19 on productivity, satisfaction, and work patterns.

Online March 17, starting at 4 p.m. Claire Tomlin “will deliver the Vanderbilt School of Engineering’s spring 2021 John R. and Donna S. Hall Engineering Lecture.” … “Her lecture—Safe Learning in Robotics—is open to the public.” [registration required]

Despite their massive size, successful deep artificial neural networks can exhibit a remarkably small gap between training and test performance. Conventional wisdom attributes small generalization error either to properties of the model family or to the regularization techniques used during training.

Through extensive systematic experiments, we show how these traditional approaches fail to explain why large neural networks generalize well in practice. Specifically, our experiments establish that state-of-the-art convolutional networks for image classification trained with stochastic gradient methods easily fit a random labeling of the training data.

It shouldn’t have to be this way, but the burden of protecting your privacy as you browse the web is your own. I do what I can in my research and consulting to advocate for systemic change in design and policy to reduce the burden on the individual user; but that kind of change is slow to fruit. Meanwhile, there are tools that you, the individual, can use to partially protect yourself against the forces of surveillance capitalism and other institutional privacy threats. I will share five that I use myself.

Vele was 16 when she embarked a slave ship in 1832 at Cameroons River in West Africa. Precillia Cozzens, 35, was registered as a slave in New Orleans in 1846. Domingos, age 6, was listed in an inventory of enslaved people at Aguiar Plantation, Brazil, in 1806.

The records of these three are among more than 750,000 of people, places, events, and sources available to search in a new open-source database called Enslaved: Peoples of the Historical Slave Trade (Enslaved.org), a repository of information and stories about those who were enslaved or enslavers, worked in the slave trade, or helped emancipate enslaved people. The entries run from the 15th century to the late 1800s and span Western Europe, Africa, and North and South America.