Large-scale DNA-based phenotypic recording and deep learning enable highly accurate sequence-function mapping

摘要

Generalized genetic architecture underlying the uASPIre approach. A genetic sequence of interest (diversifier; e.g. a GRE) controls, either positively (+) or negatively (−), the activity of a DNA-modifying enzyme (modifier), which can modify its cognate substrate DNA (discriminator). If placed on the same DNA molecule, diversifier sequence and discriminator state can be both determined by sequencing, for instance by NGS using forward (seq ) and reverse (seq ) primers. Readout of the uASPIre method. Under monoclonal conditions (i.e., only one diversifier variant per compartment/cell), sequencing of multiple DNA copies that share the same diversifier allows to determine the fraction of modified discriminators, which can be used as a continuous, normalized readout for diversifier function. Prototype plasmid employing recombinase Bxb1 as a modifier controlled by the rhamnose-inducible promoter . Bxb1 inverts an mCherry CDS into the correct orientation relative to a constitutive promoter , thus activating expression. / and / : Bxb1 attachment sites before and after recombination. , Kinetics of Bxb1-mediated discriminator modification in shake flask cultivations of . Recombination is detected by direct fluorescent measurement (open diamonds, mean and s.d. of  = 3 technical replicates) and counting of red colonies after plasmid extraction and retransformation (closed diamonds,  = 1, minimum of 275 total colonies per data point) as well as bulk Sanger sequencing of the discriminator sequence. Source data for are available as a Source Data file.