The genome editing revolution just got an upgrade. Scientists at Massachusetts General Hospital, part of the Mass General Brigham healthcare system, have developed a machine learning model called PAMmla that could reshape the way researchers design CRISPR-Cas9 enzymes for gene therapy. Published in Nature, their findings demonstrate how artificial intelligence can be used to predict, evaluate, and tailor the behavior of tens of millions of potential genome editors—an effort that could drastically reduce off-target effects and expand the therapeutic potential of CRISPR.
“Our study is a first step in dramatically expanding our repertoire of effective and safe CRISPR-Cas9 enzymes,” said Ben Kleinstiver, PhD, corresponding author of the study and Kayden-Lambert MGH Research Scholar. “In our manuscript, we demonstrate the utility of these PAMmla-predicted enzymes to precisely edit disease-causing sequences in primary human cells and in mice.”
CRISPR-Cas9 has become the go-to tool for precision gene editing, enabling scientists to cut and modify specific stretches of DNA. But the system has its shortcomings—chief among them, the risk of off-target edits that can inadvertently damage healthy genes. That’s where PAMmla comes in. Trained on an enormous dataset of enzyme activity, the algorithm can predict which enzyme variants will home in on their targets with surgical precision. Unlike earlier engineering methods, which were slower and generated a limited number of usable enzymes, PAMmla is massively scalable—capable of evaluating more than 64 million enzymes for use in therapy.
One of CRISPR-Cas9’s inherent limitations is its reliance on binding to a short DNA sequence known as the protospacer adjacent motif, or PAM. To overcome this, the researchers used PAMmla to identify Cas9 variants with novel PAM preferences that offered better specificity and functionality. These optimized enzymes were then tested in both cultured human cells and a mouse model of retinitis pigmentosa. The result? Bespoke enzymes that zeroed in on their targets with enhanced accuracy.
“A major outcome of this work is the creation of this PAMmla model that can now be used by researchers to predict customized enzymes that are uniquely tuned for their specific use cases,” said lead author Rachel A. Silverstein, a PhD candidate and NSERC postgraduate scholar. “The result of this model is that we now have an enormous toolbox of safe and precise Cas9 proteins that can be utilized for a variety of research and therapeutic applications.”
By marrying machine learning with synthetic biology, the team at MGH is helping to unlock a new era of programmable medicine—where customized gene editors could one day be designed to treat everything from inherited eye diseases to rare genetic disorders, safely and precisely.
Lorem ipsum dolor sit amet, consectetur adip elit. Donec posuere dolor massa, pellentesque aliquam nisl facilisis sed.
The genome editing revolution just got an upgrade. Scientists at Massachusetts General Hospital, part of the Mass General Brigham healthcare system, have developed a machine learning model called PAMmla that could reshape the way researchers design CRISPR-Cas9 enzymes for gene therapy. Published in Nature, their findings demonstrate how artificial intelligence can be used to predict, evaluate, and tailor the behavior of tens of millions of potential genome editors—an effort that could drastically reduce off-target effects and expand the therapeutic potential of CRISPR.
“Our study is a first step in dramatically expanding our repertoire of effective and safe CRISPR-Cas9 enzymes,” said Ben Kleinstiver, PhD, corresponding author of the study and Kayden-Lambert MGH Research Scholar. “In our manuscript, we demonstrate the utility of these PAMmla-predicted enzymes to precisely edit disease-causing sequences in primary human cells and in mice.”
CRISPR-Cas9 has become the go-to tool for precision gene editing, enabling scientists to cut and modify specific stretches of DNA. But the system has its shortcomings—chief among them, the risk of off-target edits that can inadvertently damage healthy genes. That’s where PAMmla comes in. Trained on an enormous dataset of enzyme activity, the algorithm can predict which enzyme variants will home in on their targets with surgical precision. Unlike earlier engineering methods, which were slower and generated a limited number of usable enzymes, PAMmla is massively scalable—capable of evaluating more than 64 million enzymes for use in therapy.
One of CRISPR-Cas9’s inherent limitations is its reliance on binding to a short DNA sequence known as the protospacer adjacent motif, or PAM. To overcome this, the researchers used PAMmla to identify Cas9 variants with novel PAM preferences that offered better specificity and functionality. These optimized enzymes were then tested in both cultured human cells and a mouse model of retinitis pigmentosa. The result? Bespoke enzymes that zeroed in on their targets with enhanced accuracy.
“A major outcome of this work is the creation of this PAMmla model that can now be used by researchers to predict customized enzymes that are uniquely tuned for their specific use cases,” said lead author Rachel A. Silverstein, a PhD candidate and NSERC postgraduate scholar. “The result of this model is that we now have an enormous toolbox of safe and precise Cas9 proteins that can be utilized for a variety of research and therapeutic applications.”
By marrying machine learning with synthetic biology, the team at MGH is helping to unlock a new era of programmable medicine—where customized gene editors could one day be designed to treat everything from inherited eye diseases to rare genetic disorders, safely and precisely.
Lorem ipsum dolor sit amet, consectetur adip elit. Donec posuere dolor massa, pellentesque aliquam nisl facilisis sed.
.svg)
.jpg)
.gif)