Using machine learning to super-charge anti-infective drug discovery: the case of Halicin

30212411048_2a1d7200e2_b

Yes, it’s true. There is more to HCAI & AMR (and this blog) than COVID-19! To prove it, I’m posting on something different today – the use of AI to streamline the anti-infective drug discovery process. Scientists at MIT have used machine learning (aka “deep learning”) to improve the drug discovery process, by predicting antimicrobial activity in molecules that are different from known antibiotics. This process has yielded Halicin, a promising candidate molecule for a broad-spectrum antimicrobial agent – which is, of course, a long way from clinical trials!

The team worked with libraries of known chemicals and used machine learning to predict antimicrobial activity. This identified a molecule with promising antimicrobial activity – Halicin – which had been previously developed as a non-anti-infective drug but not taken very far and abandoned. Halicin showed in vitro antimicrobial activity against a range of Gram-negatives including CPE, Pseudomonas aeruginosa and Acinetobacter baumannii. Halicin also showed promising activity in treating C. difficile and A. baumannii infections in murine models.

Halicin is a promising molecule and will proceed through to clinical trials in due course, I’m sure. But the reason for all the interest (and this paper has generated lots e.g. here and here) is the novelty of the process used. The authors illustrated the potential power of machine learning in this context by interrogating a larger molecular database than the one in which Halicin was discovered and identifying a further eight molecules with predicted antimicrobial activity that are structurally different from known antibiotics.

I gave a talk yesterday at Imperial College London Institute for Institute for Molecular Science and Engineering (IMSE) seminar about the challenges of HCAI & AMR (slides here). My Imperial colleague Dr Pau Herrero-Viñas gave a talk in the same seminar about his work around using machine learning to support diagnosis and treatment of infection. So it seems we’re seeing more and more of AI/machine learning coming through in improving our ability to identify and treat infection.

Finally, there’s a nice Inside Science BBC Radio 4 podcast on this topic this week – in addition to a social science experiment about “composting” packaging (only available to UK users, sorry).

Image: Flickr.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s