---

We infer a large fitness landscape from high-throughput sequence data from the E. coli lac promoter region with ~200k sequences. The sequences are associated with measurements of transcriptional activity. Utilizing linear regression and L1 regularization (LASSO), we find the best linear and quadratic approximations to fit the data. We find the fitness landscape to be largely smooth and additive, with a small amount of epistasis. Our method also reveals the locations of binding sites, and their interactions without any prior knowledge and without any difficult optimization steps.

---

Jakub Otwinowski, Department of Physics, Emory University
Konferenzraum Theoretische Physik
Contact: not specified