- The Aurorean
- Posts
- 🧬 Using Genomic Data To Inform Cancer Treatments | #2
🧬 Using Genomic Data To Inform Cancer Treatments | #2
+ ancient DNA, using AI to discover new battery materials and more
Hello!
Welcome back to another edition of The Aurorean.
Distinctive is the theme sprinkled throughout this week’s news and research.
Curious why? Wondering what science discovered last week? Let’s find out.
Quote of the Week 💬
NHS Uses Whole-Gene Sequencing To Tailor Cancer Care
“This study is an important milestone in genomic medicine. We are starting to realise the promise of precision oncology that was envisioned ten years ago when the 100,000 Genomes Project was launched.”
⌛ The Seven Second Summary: In the largest study of its kind, a team of scientists report how combining health data with genomic data for patients with cancer can help doctors provide more personalized care for their patients.
🔬 How It Was Done:
Clinical data and health data, such as hospital visits and the types of treatments received, were collected from 13,880 participants with 33 different types of cancers. In addition, the full genome sequence of all the participants were gathered.
The DNA from tumors and healthy tissues were examined to identify if any genetic changes in cancerous DNA led to patient outcomes with better or worse survival rates.
🧮 Key Results:
15 genes were found to affect overall survival rates across the 33 different types of cancers.
“Over 90% of brain tumors and over 50% of colon and lung cancers showed genetic changes that could affect how patients are treated, guiding decisions about surgery or specific treatments they might need.”
Furthermore, over 10% of ovarian cancers were found to have inherited genetic risks.
💡 Why This May Matter: Pairing data on specific genetic changes in cancerous DNA with information about patient outcomes and clinical health data may improve clinical decision-making by allowing patients to receive more tailored cancer treatments in the future, which may hopefully lead to better cancer survival rates over time.
🔎 Elements To Consider: There are over 100 different types of known cancers. Even though this study is the largest of its kind, there are many other types of cancerous DNA to study in order to build a more comprehensive understanding of how genomic data can best inform clinical decision-making.
🧵 Thematic Thread: Precision medicine is inherently about providing individualized and distinctive medical care based on the unique characteristics of each patient. Thus, gaining insights about the genomic data of healthy and cancerous cells in a patient is a step towards realizing this future.
📚 Learn More: East Genomics. Nature.
Stat of the Week 📊
AI & High Performance Computing Discovers New Battery Material
80 hours
⌛ The Seven Second Summary: Microsoft and the Pacific Northwest National Laboratory (PNNL) partnered to use AI and high-performance computing for a chemistry and material science project. Within a few weeks, they unearthed a promising new battery material with the potential to substantially decrease the reliance on lithium.
🔬 How It Was Done:
An AI system was trained to identify every possible element and the combinations it can be used in to conceivable accomplish the goal of the project. It proposed 32 million potential candidates.
Then, a different AI model was used to filter out materials from their initial list that were not stable. This reduced the list down to 500,000 materials.
Afterwards, another AI was used to filter out materials based on their potential to conduct energy. This reduced their list down to 800 materials.
From there, simulations were run to analyze the movements of atoms and molecules inside the remaining materials. 150 materials were still considered quality after this process.
Finally, an assessment was made of the practicality of using each material, by assessing its general availability, cost, and other factors. This left the team with 23 possible materials, 5 of which were already known to science.
🧮 Key Results: It only took 80 hours for the team to identify its 23 promising candidates from their initial list of 32 million materials. Without the use of AI, this process would have taken months or years
💡 Why This May Matter: Just last year the U.S. Department of Energy (DOE) mentioned the results of a chemistry and material science project where they utilized AI in a similar fashion as Microsoft and PNNL. The DOE mentioned AI saved their project 5 years worth of computational time, which is an incredible demonstration of how this emergent tool can significantly accelerate the pace of scientific discovery in chemistry and material science the years ahead.
🔎 Elements To Consider: Even though the initial screening process to identify suitable materials can be completed much faster now, the process to assess the long-term viability of newly discovered materials still requires a significant amount of time before they are used practically in real-world scenarios.
🧵 Thematic Thread: Identifying a short list of anything from a pool of several million is like finding a needle in a haystack. Luckily, AI can help us find the things we want with the distinctive features we are looking for efficiently.
📚 Learn More: Microsoft. arXiv. Github.
AI x Science 🤖
Credit: Alexander Sinn on Unsplash
Using AI Web Agents To Make The Internet More Accessible
In an effort to enhance internet navigation and accessibility, a research team developed a dataset for AI web agents to complete 2,350 tasks from 137 live websites. These tasks include activities like booking international flights, following celebrity accounts on social media, and scheduling tests at the DMV. The top-performing AI agents demonstrated a success rate of ~39% in completing tasks on websites and domains they had not encountered before, showcasing an evolving capability to build generalized AI agents to execute a number of tasks on its users’ behalf. The Ohio State University. arXiv. Github.
Our Full AI Index
Business: OpenAI debuts its GPT Store for users to buy and sell customized chatbots. OpenAI.
Award Winners: The FDA announced its winners of an initiative it led to develop AI models capable of predicting cardiovascular health related outcomes in Veterans. Out of 25 team entries, 5 were considered Top Performers, and a formal publication is expected to be released to provide details about the models entered in the challenge. precisionFDA.
CES 2024: There were several incredible gadgets featured at the Consumer Electronics Show this year. Some of our favorites were assistive products, such as:
Naqi’s Neural Earbuds, which is a noninvasive earbud headset that detects the wearer’s brain waves and muscle impulses and converts them into signals to control other connected bluetooth devices, such as wheelchairs, computers and IoT devices.
Gyrogear’s GyroGlove, which helps to reduce hand shaking for people with tremors.
The Whispp app, which converts the voice of someone with impaired speech patterns into a natural sounding voice.
🧵 Thematic Thread: As AI web agents become increasingly proficient at navigating the Internet and successfully executing tasks, it is possible we enter a future where individuals can deploy personalized Internet bots. These bots would be capable of performing distinctive tasks, such as gather information, and complete assignments on behalf of users, irrespective of their tech savviness or accessibility restrictions.
Other Observations 📰
Credit: Sangharsh Lohakare on Unsplash
Using Ancient DNA To Trace The Origins Of Syndromes
Two distinct teams of researchers released papers last week about their studies of ancient human DNA to identify the origins of syndromes. One team focused on researching multiple sclerosis, while the other team studied chromosomal syndromes.
The team studying multiple sclerosis analyzed DNA from individuals migrating from Asia to Europe thousands of years ago to piece together when the first genetic mutation resulting in the disease may have occurred. They aimed to infer the environmental and lifestyle conditions that may have caused the mutation to arise and persist in the people's future generations.
Meanwhile, the team studying chromosomal syndromes developed a computational method to count the number of copies of X and Y chromosomes in a dataset of ancient DNA. This method helped determine who the first individuals with a missing or extra chromosome were, the time period they are from, their geographical locations, and infer how their society may have viewed them.
Both studies provide a historical perspective on ancient human populations, enhancing our understanding of the genetic and health-related aspects of past civilizations. Continued exploration of our distant past may offer valuable insights into overcoming challenges both in the present and future.
Multiple Sclorsis Study: Washington Post. Nature.
Chromosomal Study: The Francis Crick Institute. Nature.
Our Full Science Index
Public Health: The WHO declared Cabo Verde, a country with a population of ~500,000 people, malaria-free. WHO.
Energy: The International Energy Agency (IEA) released a new report highlighting the amount of renewable energy capacity (510 gigawatts) added to energy systems around the world grew by 50% in 2023. The IEA projects under current policies and market conditions, global renewable capacity is on course to increase 2.5x by 2030, which is shy of the COP28 goal to 3x by 2030, but is still within reach if more is done. IEA.
Renewables: According to an analysis from the Institute for Energy Economics and Financial Analysis, in 2023, the U.S. generated more electricity from solar, wind, and hydro than coal on 257 days. IEEFA.
FDA: The FDA published its report of its new drug approvals in 2023. 56 drugs in total were approved last year, which is the second most yearly drug approvals seen in the last 10 years. What might get approved in 2024? FDA.
Policy: The EU announced €4bn of state aid investments in new factories producing electric batteries for cars, heat pumps and solar panels. The Guardian.
🧵 Thematic Thread: These studies in ancient DNA research aimed to uncover when distinctive traits and characteristics first emerged among groups of people, and to infer the health and societal ramifications for being the first individuals with these unique genetic features.
Media of the Week 📸
The Vulcan Rocket Launch Was A Success
United Launch Alliance successfully launched its massive 202-foot-tall (61.6-meter) Vulcan rocket on January 8th. The inaugural mission, called Cert-1, deployed payloads for NASA and Celestis, and may be the beginning of another competitive player in the space launch market. Watch its rocket launch video here! United Launch Alliance.
NASA Telescopes Capture ‘Green Monster’ In Star Debris
Credit: X-ray: NASA/CXC/SAO; Optical: NASA/ESA/STScI; IR: NASA/ESA/CSA/STScI/Milisavljevic et al, NASA/JPL/CalTech; Image Processing: NASA/CXC/SAO/J. Schmidt and K. Arcand
Astronomers utilized data from NASA's Chandra X-ray Observatory and James Webb Space Telescope to explore the supernova remnant Cassiopeia A. As they studied the blast wave of the explosion, they measured its radioactive debris, which they describe as a Green Monster. NASA.
In The Scope
Xenon nanocluster between two graphene layers, with sizes between two and ten atoms. Credit: Manuel Längle
Shifting gears from the cosmos to the microscopic. In what might be the first recorded event of its kind, researchers successfully obtained stable and direct images of clusters of gas atoms at room temperature. This breakthrough not only advances the possibilities for condensed matter physics, it also presents another avenue to study material growth and behavior. University of Vienna. Nature.
This Week In The Cosmos 🪐
No major astronomical events this week! How disappointing…
Credit: Mark Basarab on Unsplash
That’s all for this week! Thanks for reading.
Until Next Time 💭
What is your level of satisfaction with this week's newsletter? |
If you have more feedback you want to share, send us a message!