Disclosed herein are methods for the generation of highly accurate nucleic acid libraries encoding for predetermined variants of a nucleic acid sequence. The nucleic acid sequence may encode for all or part of a reference domain of a CAR. The degree of variation may be complete, resulting in a saturated variant library, or less than complete, resulting in a non-saturating library of variants. The variant nucleic acid libraries described herein may be designed for further processing by transcription or translation. The variant nucleic acid libraries described herein may be designed to generate variant RNA, DNA and/or protein populations. Further provided herein are method for identifying variant species with increased or decreased activities, with applications in regulating biological functions and the design of therapeutics for treatment or reduction of a disease, such as cancer.