Sequence design for three-dimensional genome folding using Akita Semifreddo.

Area of Science:

• Genomics
• Molecular Biology
• Bioinformatics

Background:

• Mammalian genomes possess intricate 3D structures vital for gene regulation.
• Local genome organization features like boundaries, dots, and flames are driven by loop extrusion.
• Designing DNA sequences for specific 3D folding patterns remains a challenge.

Purpose of the Study:

• To present Akita Semifreddo, a framework for the rational *in silico* design of DNA sequences with programmable 3D folding outcomes.
• To demonstrate the framework's ability to generate diverse local folding features.
• To explore the relationship between DNA sequence and chromatin architecture.

Main Methods:

• Combined a computationally efficient "half-frozen" AkitaV2 genome folding model with the Ledidi sequence optimizer.
• Systematically designed synthetic DNA sequences (∼2 kb) to induce specific local folding features.
• Investigated the role of CTCF motifs and transcription-associated features in boundary formation and suppression.

Main Results:

• Akita Semifreddo successfully generated sequences inducing boundaries, dots, and flames with controllable strengths.
• CTCF motif configurations were consistent with known mechanisms for boundary formation.
• Weak boundaries were achieved using transcription-associated features, and strong boundaries were suppressed by SINE B2 retroelement-like sequences.
• Revealed a many-to-one relationship between DNA sequence and 3D folding outcomes.

Conclusions:

• Akita Semifreddo enables the rational design of DNA sequences for predictable 3D genome organization.
• The study uncovers the sequence grammar underlying 3D chromatin architecture.
• Provides a platform for engineering synthetic regulatory landscapes and dissecting chromatin folding mechanisms.