Almost immediately after the coronavirus emerged in the spotlight, theories surfaced suggesting it might have been intentionally created through experiments in various Wuhan laboratories. Some politicians fueled this narrative, prompting the White House to urge intelligence agencies to investigate potential lab connections.

The scientific community largely supports the view that the virus likely jumped from animals to humans. On April 30, the U.S. Office of the Director of National Intelligence released a statement on behalf of its 17 agencies, asserting that "the Covid-19 virus was not manmade or genetically modified." They opted to further explore two possibilities: the prevailing theory of zoonotic transmission and the less probable scenario of a natural virus accidentally released from a lab.

The U.S. intelligence community aligns with this broad scientific agreement, affirming that the virus wasn’t human-made. Yet, how did they reach this conclusion? While the entire investigation remains undisclosed, one intelligence initiative named FELIX specifically examined this hypothesis. Their findings indicated that the virus did not utilize "foreign" genetic sequences, suggesting that SARS-CoV-2, responsible for Covid-19, is not artificially engineered.

Identifying "bioengineering" presents challenges for any organism. Multiple methods exist for determining if a virus is engineered, but there are also numerous ways to engineer a virus, resulting in ongoing debate and considerable uncertainty.

FELIX, which stands for Finding Engineering-Linked Indicators, operates under IARPA (Intelligence Advanced Research Projects Activity). This initiative engages in high-risk research and advances next-generation technologies. In 2018, FELIX began funding six external teams to develop tools capable of identifying the markers of bioengineering—genetic indicators revealing manipulations of an organism's genome.

A genome encompasses the complete array of genetic bases constituting an organism. In DNA, these bases are A, G, C, and T, while in RNA, they are A, G, C, and U. When assembled, they form "sequences," which can denote either the entire set of genetic letters in a specified order or a smaller segment.

According to FELIX program manager Dr. David Markowitz, indicators of engineering within a genome can manifest in various ways. These include the presence of foreign genetic material in a sequence or irregular duplications, insertions, or deletions of bases. Other indicators, as noted by Dr. Isaac Plant, a former FELIX graduate student at Harvard, consist of sequences associated with antibiotic resistance and short segments termed "scars," indicating alterations to a DNA sequence. While no exhaustive list of "engineering sequences" exists, resources like Addgene maintain extensive databases for molecular manipulation tools.

FELIX's tools could identify instances where biological intellectual property has been misappropriated, such as a unique yeast strain appearing in a competitor's lab, as well as assess the naturalness of emerging pathogens. The initiative faced its first substantial real-world test with SARS-CoV-2, an RNA virus.

“FELIX teams dedicated 18 months to developing initial prototypes of their detection platforms,” stated Markowitz. “They were ready to tackle SARS-CoV-2 when this biosecurity threat emerged.”

In January, a team from the MIT-Broad Foundry utilized FELIX tools to evaluate claims regarding the lab-engineered origins of SARS-CoV-2, as per IARPA’s website. Although the results are not fully disclosed, the system compared the virus's genome against 58 million known genetic sequences, including those of closely and distantly related viruses. In just ten minutes, the tool determined that the virus's genetic structure aligned more closely with naturally occurring coronaviruses than with any other organisms, concluding that "no sequences from foreign species have been engineered into SARS-CoV-2," according to IARPA.

While this conclusion appears definitive, Dr. Filippa Lentzos, a senior research fellow at King’s College London focused on biosecurity, points out that it does not entirely exclude the possibility of engineering; it merely suggests that the virus wasn’t engineered in specific ways.

The MIT-Broad team involved with SARS-CoV-2 declined an interview, but other FELIX-funded teams were more forthcoming. Dr. Eric Young, who studies yeast engineering at Worcester Polytechnic Institute, turned his attention to biosecurity after observing policy-makers and ethicists at synthetic biology conferences expressing concerns about the implications of easily creating custom organisms.

“What we’re developing holds great promise for civilization,” Young stated, emphasizing advancements such as new pharmaceuticals. "However, history shows that technological progress can be repurposed for harmful uses." He noted that, to date, there have been no recorded instances of synthetic biology being used to create bioengineered weapons.

The tool Young worked on for FELIX initially sequences an organism’s entire genome and returns the findings to the researcher with annotations highlighting any genetic components that appear altered. It checks for signs of engineering by comparing the sequence against a standard list of known yeast engineering sequences or a custom list provided by the user. Future iterations will leverage machine learning to identify engineered DNA sequences without needing predefined lists, although current functionality is limited to yeast sequences.

Ginkgo Bioworks, another recipient of FELIX funding, is tackling the challenge with computational biology and a proprietary database of known engineered sequences. This company frequently engages in designing and engineering organisms. "For this program, we generated over 6 million simulated, engineered genomes reflecting diverse genetic engineering techniques across organisms chosen by IARPA and national labs," stated Dr. Joshua Dunn, Ginkgo’s head of design.

“We aimed to be the ethical hackers in this scenario,” Dunn explained. To achieve this, Ginkgo employs several approaches: first, they compare a sequence to known natural reference sequences, similar to the MIT-Broad Foundry's work on SARS-CoV-2. Next, they search for established engineering signatures within the same sequence. A third method analyzes the genetic alphabet's distribution. Finally, they are developing a fusion engine that integrates these results to provide a comprehensive assessment.

At Draper Laboratory, led by Dr. Kirsty McFarland, a FELIX group is focusing on detecting engineered organisms in environmental samples teeming with life, such as soil or water. They are devising distinct methods aimed at identifying two types of bioengineering: unknown alterations in known organisms and recognized or suspected changes in potentially unknown organisms. SARS-CoV-2, as a novel pathogen, fits the latter category.

The first method involves comparing discovered genomes to a database of anticipated gene sequences. The second seeks specific sequences that may indicate engineering within the genomes of previously unidentified organisms. Currently, this second method can detect a single engineered organism even among a million or more naturally occurring entities.

At Harvard, Dr. Elizabeth Libby leads a team developing a "biosensor," an engineered cell capable of detecting other engineered organisms. The Boston group initially conceptualized the sensor on a Dunkin' Donuts napkin. “It can essentially absorb any nearby DNA,” Libby explained. It then identifies preprogrammed DNA signatures, amplifies the signal, and lights up to indicate detection.

For FELIX, they have programmed the biosensor to respond to common engineering markers, but the technology has broader applications. "In the future, we envision a disposable cartridge that passively senses for pathogens in environments like air-handling systems, hospital surfaces, or water supplies," Libby noted. If it detects "the thing of interest"—be it an engineering marker, a coronavirus, or another genetic sequence—the cartridge illuminates and sends a signal. In the context of a pandemic, such a device could be invaluable for determining the presence of pathogens in public spaces.

Currently, these methodologies largely share a common limitation: they depend on records of known organisms or engineering signatures. Essentially, they require a reference catalog for comparison. The statement on IARPA's website suggests that the MIT-Broad Foundry’s analysis does, too.

“You will never have a foolproof system for detecting engineered organisms,” cautioned Plant.

This encapsulates the challenge: Presently, most analyses hinge on existing data and assumptions about the multifaceted nature of engineering. “Those attempting to engineer organisms are just as knowledgeable as those trying to detect such changes,” Plant remarked, leading to what can be described as an arms race. “Complete detection of engineered organisms is an impossible task.”

The MIT-Broad Foundry's analysis of SARS-CoV-2 dismissed the notion that the virus was constructed from parts of other organisms—a common technique among FELIX teams. Yet, many pathogens possess genomes not represented in existing databases. “You can't account for all viruses that have been identified but remain unpublished,” noted Dr. Alina Chan, a postdoctoral researcher at the Broad Institute. Many scientists take years to publish their findings, and there are no regulations prohibiting data hoarding.

Although such delays are typical, they hinder initiatives like FELIX, as some detection tools function similarly to plagiarism detectors, checking for instances of unauthorized use of genetic material. A lazy plagiarist might copy text verbatim—easily identifiable. A more skilled plagiarist may paraphrase, complicating detection. A "super detector" might recognize whether a work is influenced by another source. All plagiarism detectors, akin to engineering detectors, require a robust catalog of published works. If someone manipulates an unpublished virus sourced from the wild, detection becomes significantly more challenging.

Plant offers a different analogy for detecting engineered organisms. “Identifying a manipulated organism is akin to determining whether a new word has been deliberately invented,” he explained. "Achieving perfect confidence in this requires knowing every word that has ever existed, as well as all words that might be formed by accident. This, like flawlessly detecting engineered organisms, is unfeasible.” Efforts like FELIX may never be fully equipped to ascertain whether an organism is engineered.

FELIX's current analysis of SARS-CoV-2 could identify instances where scientists incorporated publicly available sequences into pathogens. However, a more astute bioengineer might employ subtler methods. “If you wanted to blend into an event unnoticed, you would disguise yourself as an unknown individual,” Chan elaborated, “rather than impersonating multiple well-known celebrities.”

What FELIX has demonstrated is that SARS-CoV-2 isn't a composite of various well-known genetic components, which is a valuable hypothesis to eliminate. “This was a limited application of a toolkit,” remarked Dr. Gregory Koblentz, associate professor and director of the biodefense graduate program at George Mason University.

Even if FELIX perfected its genetic engineering detection capabilities, its findings would likely raise more questions than answers. In the event of a disease outbreak, intelligence and health officials would still seek to know where the germ was engineered, who was responsible, and the rationale behind it. “This necessitates a deeper understanding,” Koblentz emphasized—not just technical knowledge, but also intelligence and law enforcement insights.

FELIX doesn't directly address these queries, yet it raises alternative considerations. While its primary goal is to enhance biosecurity, its technology is inherently dual-use, presenting both offensive and defensive capabilities. Understanding how to detect bioengineering implicitly provides insights into how to conceal it. In this manner, FELIX could be perceived as a subtle display of strength to the international community, showcasing offensive capabilities without breaching bioweapons treaties. IARPA did not respond to inquiries regarding dual-use implications prior to publication.

Programs like FELIX convey a broader message to the world. “This research stems from the belief that the spread of increasingly advanced biotechnology creates new threats we are unprepared to identify,” Koblentz stated. “This initiative aims to prevent future surprises akin to Pearl Harbor.” FELIX is merely one example of such efforts.

IARPA and DARPA have various similar programs in place.

In essence, these initiatives signal to the world that the U.S. perceives biothreats as immediate dangers. This implication, according to Lentzos, could prompt other nations to pursue more pathogenic research in order to avoid being outpaced. “While you’re merely trying to safeguard your own interests,” she cautioned, “you may inadvertently be cultivating a greater threat.”