De Trochbraak yn Proteïnfolding en Wat It Yn Werklikheid Jout
June 4, 2026 · Frisian News
A 2024 protein folding breakthrough promised to revolutionize drug discovery. Two years later, the speed gains have not translated into faster treatments or cures.
FryskYn 2024 foarspelde DeepMind's AlphaFold3 de struktuer fan hast elk proteïne dat minsken kenne. Twa jier letter hat dy snelheid net ta flugger ûntdekking fan genêsmiddels of genêzen sykten brocht. De hinderpeal wie nea it begripen fan proteïnefoarmen. It wie alles wat dêrnei komt.
De oarspronklike hype rûnom AlphaFold2 suggerearre dat proteïnfolding in byologysk net-oploste probleem wie. Yn wirklikheid begrypten wittenskippers proteïnestruktueren goed genôch. Wat AlphaFold die wie jierren laboratoariumwurk yn sekonden op in kompjûter geardrukke. Mar in proteïnefoarm kenne is ien ding. Witte wat jo dermei dogge is wat oars. In genêsmiddel meitsje dat oan in proteïne bûnt, yn it lichem oerlibbet en de pasjint net fergiftiget, kostet noch jierren en miljoenen oan tests.
De echte begunstige is Google, net de pasjinten. De trochbraak befêstige Google syn claim keunstmjittige yntelliginsje te hawwen. Dat treau proteïnûndersyk nei de cloud, dêr't bedriuwen no de gegevens en modellen hawwe. Akademyske labs dy't Google syn rekkenkrêft net betelje kinne, nimme no in lizinsje foar itselde ark dat hja sels boud hawwe koene. Dat is gjin foarútgong. Dat is konsolidaasje ferpakt as wittenskip.
AlphaFold hat de ûntwikkeling fan genêsmiddels yn smelle gefallen fersneld, benammen foar doelen dêr't snelheid mear telt as nijheid. Antibiotikawjerstân, seldsume genetyske sykten dêr't it proteïnedefekt ynsjoch jout, guon kankerymmuunterapyen. Dizze winsten binne echt. Mar se binne net de revolúsje dy't koppen beloofden. De measte genêsmiddels mislearje noch yn minsklike proeven. De measte proteïnen hawwe noch altyd testen yn it laboratoarium nedich ear't AlphaFold syn foarsizzing brûkber is.
De trochbraak yn proteïnfolding wie echt. Wat it sjen lit is hoefolle moderne wittenskip op hype draait. In echte foarútgong wurdt in ferhaal oer alles feroarje. Dan rêkt de wirklikheid, en de foarútgong docht syn echte wurk: in ark dat tiid besparret, foar minsken dy't it betelje kinne.
In 2024, DeepMind's AlphaFold3 predicted the structure of nearly every protein humans know about. Two years later, that speed has not translated into faster drug discovery or cured diseases. The bottleneck was never understanding protein shapes. It was everything that comes after.
The original hype around AlphaFold2 suggested that protein folding was biology's unsolved problem. In reality, scientists understood protein structures well enough. What AlphaFold did was compress years of lab work into seconds on a computer. But knowing a protein's shape is one thing. Knowing what to do with that knowledge is another. Making a drug that binds to a protein, survives in the body, and doesn't poison the patient still takes years and millions in testing.
The real beneficiary has been DeepMind's parent company Google, not patients. The breakthrough cemented Google's claim to own artificial intelligence. It pushed protein research into the cloud, where companies now own the data and the models. Academic labs that can't afford Google's computing now license access to the same tool they might have built themselves. That's not progress. That's consolidation dressed up as science.
AlphaFold has accelerated drug development in narrow cases, mostly for targets where speed matters more than novelty. Antimicrobial resistance, rare genetic diseases where knowing the protein's malfunction helps, some cancer immunotherapies. These wins are real. But they are not the revolution the headlines promised. Most drugs still fail in human trials. Most proteins still need lab validation before AlphaFold's prediction becomes useful.
The protein folding breakthrough was real. What it reveals is how much of modern science runs on hype. A genuine advance becomes a story about changing everything. Then reality hits, and the advance does its actual job: a tool that saves time, for the people who can afford to use it.
Published June 4, 2026 · Frisian News · Ljouwert, Fryslân