In proteins consistently decrease, while the frequencies of Ser, His, Cys, Met and Phe increase over the course of protein evolution. The trend of amino acid gain and loss is in agreement with the likely order of incorporation of amino acids into the genetic code, as deduced from other criteria. Several protein design experiments have proved that the full set of 20 amino acids is not necessarily essential for protein structure and function. Further, Hecht group created four helix bundle proteins with 11 amino acids and Jumawid et al. generated a3b3 de novo proteins with seven amino acids. However, these experiments have attempted to generate simplified proteins with fewer amino acids than the natural proteins, and they have not focused on whether the accepted amino acids are primitive or not. Previously, Babajide et al. demonstrated in silico that native-like folded structures of several tested proteins are maintained with a restricted GANT61 cost alphabet mainly containing primitive amino acids but were not maintained with a set of nonprimitive amino acids. To test this hypothesis experimentally, we sought to compare the function and structure of tested proteins with different subsets of amino acids for the first time. As a first attempt, we designed randomized src SH3 gene libraries in which approximately half the residues of the SH3 gene were replaced by randomized codons in the lower or upper half of the table of the genetic code. The SH3 domain is one of the most common mediators in intracellular signaling pathways. Because the SH3 domain is a well-known protein and thus the conserved positions that play important roles in structure and function have already been examined, we can randomize only non-conserved regions. A subset of amino acids that are coded by the lower half of the genetic code are mainly putative primitive amino acids, whereas a subset of amino acids that are coded by the upper half contains many putative new amino acids. From these randomized libraries, functional SH3 sequences were selected using mRNA display. In mRNA display, each cell-free translated polypeptide in a library covalently binds to its corresponding mRNA through puromycin. After affinity selection via the protein portion of an mRNA-displayed protein library, selected proteins can be easily identified by amplification and sequencing of the mRNA portion. Moreover, mRNA display based on cell-free translation can handle larger number of molecules than the other cell-based display technique such as phage display, and it makes possible enrichment of active sequences with low abundance from a library with high diversity and complexity. Therefore, we used mRNA display to elucidate and compare the frequency of functional SH3 sequences in randomized SH3 libraries with different sets of amino acids.