AlphaFold AI predicts the structure of nearly every known protein

The incredible AI offers itself a second major scientific revolution in the space of barely a year.

In July 2021, almost a year ago to the day, DeepMind teams dropped a veritable atomic bomb in the world of molecular biology; the firm presented an incredible database where its AlphaFold AI predicted the 3D structure of almost all proteins in the human body.

If it was unanimously hailed by the entire scientific community, it was because it was a huge step forward for fundamental research; this production already promises to have a decisive impact at all levels of biology and life sciences. You can consult our complete file which summarizes the ins and outs of this tremendous progress.

And yet, it seems that the wizards of DeepMind were only at their first attempt. They have just triggered a second major academic earthquake in the space of a year by presenting a new version of their database. Last year’s had almost every protein in the human body; this one, on the other hand, predicts the structure of… almost all of the proteins documented by science.

A research tool that is already revolutionizing biology

As a reminder, the previous version has already begun to upset structural biochemistry to its very foundations in the space of a few months. And this is not an exaggeration. It is no coincidence that leading researchers have even compared it to the sequencing of the human genome, one of the most significant achievements of modern science.

If the specialists are so rave about it, it is because the problem of protein structure has been a overly tedious puzzle for researchers for decades (see the second paragraph of our dossier).

These are the elements from which life is built; therefore, they largely condition what happens within all living organisms. Their study is therefore of paramount importance in a host of areas. Much of what we know about how our bodies work comes from our knowledge of proteins.

However, their three-dimensional structure is intimately linked to their physiological role. Each discovery on their structure can therefore bring its share of very exciting novelties. But the traditional method that has been used to study them for decades is excessively slow and tedious, and by extension not very productive.

AlphaFold: Here’s why the protein fold database is revolutionary

The world of proteins will soon have no more secrets

The arrival of algorithmic tools based on AI, starting with the famous AlphaFold, has completely changed the situation in medicine. Now, you no longer need to spend months studying a protein with no guarantee of results; researchers can access AI-predicted structural information with just a few clicks.

What transform their way of working. ” AlphaFold has already accelerated and enabled massive discoveries “, explains Eric Topol, director of a biology laboratory whose daily operation has completely changed with AI.

And now, with this new global database that describes the expected structure of over 200 million proteins, that conclusion is no longer limited to the human body; it now applies to almost all of the physiological mechanisms of living organisms!

The small circles represent the amount of protein in the previous version of the database. Large colored discs show its current condition. © DeepMind

This changes a lot of things in the small world of structural biology; this is a huge step forward, the magnitude of which can easily be seen on this graph. To compare with transport, it’s a bit as if DeepMind had just invented the wheel, then aviation one after the other while humanity was still trudging barefoot a year earlier.

In other words, the scope of this work is absolutely immense. “ With this addition of structures that shed light on almost the entire protein universe, we can expect new biological mysteries to be solved every day. “Summarizes Topol. Suffice to say that we are probably at the dawn of a whole new era of research in fundamental biology!

Leave a Comment