Artificial Intelligence

Quantum-Inspired AI model helps CRISPR Cas9 Genome Editing for Microbes


A team of scientists at the Oak Ridge National Laboratory (ORNL), have embarked on a groundbreaking venture, leveraging quantum biology, artificial intelligence (AI), and bioengineering to revolutionize the effectiveness of CRISPR Cas9 genome editing tools. Their focus is on microbes, particularly those earmarked for modifications to produce renewable fuels and chemicals, presenting a unique challenge due to their distinct chromosomal structures and sizes.

Traditionally, CRISPR tools have been tailored for mammalian cells and model species, resulting in weak and inconsistent efficiency when applied to microbes. Recognizing this limitation, Carrie Eckert, leader of the Synthetic Biology group at ORNL, remarked, “Few have been geared towards microbes where the chromosomal structures and sizes are very different.” This realization prompted the ORNL scientists to explore a new frontier in the quest to enhance the precision of CRISPR tools.

The team’s journey took an unconventional turn as they delved into quantum biology, a field at the intersection of molecular biology and quantum chemistry. Quantum biology explores the influence of electronic structure on the chemical properties and interactions of nucleotides, the fundamental building blocks of DNA and RNA, within cell nuclei where genetic material resides.

To improve the modeling and design of guide RNA for CRISPR Cas9, the scientists developed an explainable AI model named the iterative random forest. Trained on a dataset of approximately 50,000 guide RNAs targeting the genome of E. coli bacteria, the model took into account quantum chemical properties. The objective was to understand, at a fundamental level, the electronic distribution in nucleotides, which influences the reactivity and stability of the Cas9 enzyme-guide RNA complex.

“The model helped us identify clues about the molecular mechanisms that underpin the efficiency of our guide RNAs,” explained Erica Prates, a computational systems biologist at ORNL. The iterative random forest, with its thousands of features and iterative nature, was trained using the high-performance Summit supercomputer at ORNL’s Oak Ridge Leadership Computer Facility.

What sets this approach apart is its commitment to explainable AI. Rather than relying on a “black box” algorithm that lacks interpretability, the ORNL team aimed to understand the biological mechanisms driving results. Jaclyn Noshay, a former ORNL computational systems biologist and first author on the paper, emphasized, “We wanted to improve our understanding of guide design rules for optimal cutting efficiency with a microbial species focus.”

Graphical Abstract https://academic.oup.com/nar/article/51/19/10147/7279034

Validation of the explainable AI model involved CRISPR Cas9 cutting experiments on E. coli, using a large group of guides selected by the model. The results were promising, confirming the efficacy of the model in guiding genome modifications for microbes.

The implications of this research extend far beyond microbial genome editing. “If you’re looking at any sort of drug development, for instance, where you’re using CRISPR to target a specific region of the genome, you must have the most accurate model to predict those guides,” highlighted Carrie Eckert. The study not only advances the field of synthetic biology but also has broader applications in drug development and bioenergy research.

The ORNL researchers envision collaborative efforts with computational science colleagues to further enhance the microbial CRISPR Cas9 model using additional data from lab experiments and diverse microbial species. The ultimate goal is to refine CRISPR Cas9 models for a wide range of species, facilitating predictive DNA modifications with unprecedented precision.

The study, supported by the DOE Office of Science Biological and Environmental Research Program, ORNL’s Lab-Directed Research and Development program, and high-performance computing resources, signifies a significant leap forward in the quest to improve CRISPR technology. As Paul Abraham, a bioanalytical chemist at ORNL, remarked, “A major goal of our research is to improve the ability to predictively modify the DNA of more organisms using CRISPR tools. This study represents an exciting advancement toward understanding how we can avoid making costly ‘typos’ in an organism’s genetic code.” The findings hold promise for applications in fields ranging from bioenergy feedstock enhancement to drug development, marking a pivotal moment in the evolution of CRISPR technology.

Sources

https://doi.org/10.1093/nar/gkad736

Kevin Ku Unsplash

The AI Paradox: Enhancing Cybersecurity while Raising Concerns

In today’s interconnected and digitized world, numerous cybersecurity attacks such as malware attacks, phishing attacks, and data breaches occur, increasing the demand for cybersecurity professionals and advanced technologies such as Artificial Intelligence. However, AI acts as a defender and as a challenger to cybersecurity. 

Cybersecurity involves analyzing patterns from massive amounts of data to protect systems and networks from digital attacks and identify threats. The traditional approach relied heavily on signature-based detection systems, which were effective against known threats but incapable of detecting new or unknown threats, resulting in frequent cybersecurity attacks. However, AI-based solutions use machine learning algorithms that are trained using vast amounts of historical and current data to detect and respond to unknown and new threats. According to Jon Olstik’s Artificial Intelligence and Cybersecurity, approximately “27 percent” of security professionals in 2018 wanted to use AI for “improving operations, prioritizing the right incidents, and even automating remediation tasks,” implying that the reasons for AI in cybersecurity continue to increase over time.

However, there’s another side to AI’s role in cybersecurity, and it’s more dangerous than beneficial. While it may appear that AI systems are unlikely to be hacked, this is not the case because hackers can manipulate these systems. As previously said, AI solutions employ machine learning algorithms to detect threats; nevertheless, a fundamental flaw is that vulnerability is that these models are fully dependent on the dataset. A poisoning attack occurs when an attacker modifies the dataset to fulfill their malicious goals, forcing the model to learn from the modified data and leading to more attacks. 

As these rising concerns and vulnerabilities, AI-based solutions will need to be not only fast, but also safe and risk-free. AI and Cybersecurity share a complex relationship and it’s safe to conclude that Artificial Intelligence will continue to play a paradoxical yet essential role in the growing field of cybersecurity.

  • Comiter, Marcus, et al. “Attacking Artificial Intelligence: AI’s Security Vulnerability and What Policymakers Can Do About It.” Belfer Center, August 2019, https://www.belfercenter.org/publication/AttackingAI#toc-3-0-0. Accessed 7 July 2023.
  • Moisset, Sonya. “How Security Analysts Can Use AI in Cybersecurity.” freeCodeCamp, 24 May 2023, https://www.freecodecamp.org/news/how-to-use-artificial-intelligence-in -cybersecurity/. Accessed 7 July 2023.
  • Oltsik, Jon. “Artificial intelligence and cybersecurity: The real deal.” CSO Online, 25 January 2018, https://www.csoonline.com/article/564385/artificial-intelligence-and -cybersecurity-the-re al-deal.html. Accessed 7 July 2023. 

Who Would You Trust More: AI or Doctors?

For as long as the profession existed, doctors have been working diligently to perfect their craft and refine any rough edges, diagnosing, treating, and eventually curing their patients in the most efficient way possible in their eyes. However, mistakes are frequently made: medical malpractice is the third leading cause of death in the United States, with over 250,000 deaths occurring yearly. Despite the rigorous education doctors undergo to officially practice their craft, they too still make mistakes. It’s human nature to err sometimes, even in life-or-death scenarios. For the majority of time, it appeared as if this was just a sacrifice that had to be made to keep one of the world’s oldest, and most vital, professions stable. 

But what if the risk of human error was eliminated by having humans removed from the equation when it came to distributing medical care?  This would dynamically pivot the medical industry and the person-to-person interaction we all know today, in a completely different direction. Some speculate that this is possible, through the utilization of artificial intelligence (AI). 

Artificial intelligence has permeated throughout the medical field briefly, but it’s been shut down due to a variety of complications, whether it’d be availability, cost, unreliability, or a combination of these factors (among others). This was especially true of Mycin, an expert system designed by Stanford University researchers to assist physicians in detecting and curing bacterial diseases. Despite its superb accuracy, being even as reliable as human experts on the matter, it was far too rigid and costly to be maintained. Despite not being medically affiliated, Google image software is another example of just how unreliable AI is: it assessed, with 100% certainty, that a slightly changed image of a cat is guacamole, a completely incorrect observation.

However, as modern technology rapidly advances, with special emphasis on machine learning (the ability of a machine to function and improve upon itself without human intervention), some believe that AI can now pick up the slack of physicians. 

This claim isn’t entirely unsubstantiated: artificial intelligence can already assess whether or not infants have certain conditions (of which there are thousands of) by facial markers, something doctors struggle with due to the massive variety of illnesses. MGene, an app that has Ai examine a photo taken of a child by its user, has over a 90% success rate at accurately detecting four serious, potentially life-threatening syndromes (Down, DiGeorge, Williams, and Noonan). AI even detected COVID-19, or SARS-CoV-2, within Wuhan, China (the origin of this virus) a week before the World Health Organization (WHO) announced it as a new virus.

With every passing day, it appears that more and more boxes that are needing to be checked, enabling the possibility of artificial intelligence becoming a dominating presence within the medical field to become one step closer to turning into a reality.

That isn’t to say that there are issues with having artificial intelligence enter the medical industry: beyond the previous problems (of cost and unreliability) being possible, Ai being ever-changing also opens up the doors to bias, ranging from socioeconomic status to race to gender and everything in between. In addition, the usage of AI also is uncomfortable to many due to the removal of the person-to-person interaction that is commonly known to people, another big issue that needs to be addressed to ensure the successful implementation of artificial intelligence into the healthcare sector. 

Regardless of what side you are on, there is a common ground: artificial intelligence will continue to get more and more advanced. While it is uncertain as to whether the general public will want AI to replace doctors, have them serve as back-end helpers, or not exist whatsoever in the office, it is clear that artificial intelligence is a tool that has both a lot of benefits and drawbacks. Whether AI is implemented or not is a question that is left to the future. 

AI can now use the help of CRISPR to precisely control gene expressions in RNA

Almost all infectious and deadly viruses are caused due to their RNA coding. Researchers from established research universities, such as NYU and Columbia, alongside the New York Genome Center, have researched and discovered a new type of CRISPR technology that targets this RNA and might just prevent the spread of deadly diseases and infections.

A new study from Nature Biotechnology has shown that the development of major gene editing tools like CRISPR will serve to be beneficial at an even larger scale. CRISPR, in a nutshell, is a gene editing piece of technology that can be used to switch gene expression on and off. Up until now, it was only known that CRISPR, with the help of the enzyme Cas9, could only edit DNA. With the recent discovery of Cas13, RNA editing might just become possible as well.

https://theconversation.com/three-ways-rna-is-being-used-in-the-next-generation-of-medical-treatment-158190

RNA is a second type of genetic material present within our cells and body, which plays an essential role in various biological roles such as regulation, expression, coding, and even decoding genes. It plays a significant role in biological processes such as protein synthesis, and these proteins are necessary to carry out various processes. 

RNA viruses

RNA viruses usually exist in 2 types – single-stranded RNA (ssRNA), and double-stranded RNA (dsRNA). RNA viruses are notoriously famous for causing the most common and the most well-known infections – examples being the common cold, influenza, Dengue, hepatitis, Ebola, and even COVID-19. These dangerous and possibly life-threatening viruses only have RNA as their genetic material. So, how can/might AI and CRISPR technology, using the enzyme Cas13 help fight against these nuisances?

Role of CRISPR-Cas13

RNA targeting CRISPRs have various applications – from editing and blocking genes to finding out possible drugs to cure said pathogenic disease/infection. As a report from NYU states, “Researchers at NYU and the New York Genome Center created a platform for RNA-targeting CRISPR screens using Cas13 to better understand RNA regulation and to identify the function of non-coding RNAs. Because RNA is the main genetic material in viruses including SARS-CoV-2 and flu,” the applications of CRISPR-Cas13 can promise us cures and newer ways to treat severe viral infections.

“Similar to DNA-targeting CRISPRs such as Cas9, we anticipate that RNA-targeting CRISPRs such as Cas13 will have an outsized impact in molecular biology and biomedical applications in the coming years,” said Neville Sanjana, associate professor of biology at NYU, associate professor of neuroscience and physiology at NYU Grossman School of Medicine. Learn more about CRISPR, Cas9, and Cas13 here

Role of AI

Artificial intelligence is becoming more and more reliant as days pass by. So much so, that it can be used to precisely target RNA coding, especially in the given case scenario. TIGER (Targeted Inhibition of Gene Expression via guide RNA design), was trained on the data from the CRISPR screens. Comparing the predictions generated by the model and laboratory tests in human cells, TIGER was able to predict both on-target and off-target activity, outperforming previous models developed for Cas13 

With the assistance of AI with an RNA-targeting CRISPR screen, TIGER’s predictions might just initiate new and more developed methods of RNA-targeting therapies. In a nutshell, AI will be able to “sieve” out undesired off-target CRISPR activity, making it a more precise and reliable method.