On the ORCID tool I worry that this would potentially lead to actually more false positives for lots of synthetic biologists and protein engineers—most only use E. coli but regularly order and express genes from a wide variety of organisms in this system (the origin of these genes is rarely publicly accesible after publication and would be very hard to map and would change with every project) I personally only work with one or two bacterial species but have ordered genes from over 100
yeah, we assume this kind of tool would be used as one part of a second-stage follow-up screen when an order has been flagged and can’t be approved by default. I agree that ordering genes from a wide variety of organisms isn’t an anomalous or suspicious signal on its own. Right now, questions like “does this person have research history with this organism?”, which some of signals of legitimacy assembled to decide whether to fulfill a flagged order, are answered by ad-hoc googling by customer service reps; tools like Cliver can automated and accelerate that process (see their recent preprint on Evaluating AI-Assisted Customer Verification for Synthetic Nucleic Acid Screening).
On the ORCID tool I worry that this would potentially lead to actually more false positives for lots of synthetic biologists and protein engineers—most only use E. coli but regularly order and express genes from a wide variety of organisms in this system (the origin of these genes is rarely publicly accesible after publication and would be very hard to map and would change with every project) I personally only work with one or two bacterial species but have ordered genes from over 100
yeah, we assume this kind of tool would be used as one part of a second-stage follow-up screen when an order has been flagged and can’t be approved by default. I agree that ordering genes from a wide variety of organisms isn’t an anomalous or suspicious signal on its own. Right now, questions like “does this person have research history with this organism?”, which some of signals of legitimacy assembled to decide whether to fulfill a flagged order, are answered by ad-hoc googling by customer service reps; tools like Cliver can automated and accelerate that process (see their recent preprint on Evaluating AI-Assisted Customer Verification for Synthetic Nucleic Acid Screening).