An ultra-high-throughput method for measuring biomolecular activities

Abstract

Methods of measuring the activities of biomolecules and pathways are fundamental to the study and engineering of biological systems. Traditional marker genes that generate light or color are limited in their dynamic range and often require expensive equipment. Here we describe a way to simultaneously detect the identity and measure the activity of each molecular variant within a mixed population of cells using nanopore sequencing. By linking the desired activity to expression of a base editor that induces mutations in a “canvas” adjacent to the DNA encoding the molecule or pathway of interest, we show that a single long-read sequencing step can retrieve identity and activity at numerous timepoints in bacteria, potentially greatly increasing the available dynamic range. Our technique can replace flow cytometry in directed evolution and may be capable of directly mapping biomolecular fitness landscapes.

FLIGHTED: Inferring Fitness Landscapes from Noisy High-Throughput Experimental Data

“…FLIGHTED applies Bayesian modeling to the problem of inferring fitness landscapes from noisy high-throughput experimental data. Practitioners must train a version of FLIGHTED for each class of high-throughput experiment with similar sources of noise; in this paper, we train two versions of FLIGHTED for single-step selection experiments and for DHARMA, a molecular recording assay that links fitness to base editing mutations in a canvas [16]. For each class of high-throughput experiment, FLIGHTED requires a calibration dataset of noisy experimental results mapped to ground-truth fitnesses; these ground-truth fitnesses can be obtained by simulation or by many repeated experimental measurements.…”

“…We next turn to DHARMA (Direct High-throughput Activity Recording and Measurement Assay), a recently developed high-throughput protein fitness assay [16]. DHARMA measures fitness by linking protein fitness to transcription of a base editor; this base editor is targeted to a canvas where it randomly causes C → T edits, as shown in Figure 3a.…”