The codon frustration index as a new metric for mRNA stability, translation efficiency, and rates of protein synthesis

Taking the human genome as a case of study, we propose a new classification of codons based only on two genomic information. We use the relative synonymous codon usage (RSCU) as a measure of non-uniform usage of synonymous codons. Similarly, we introduce here the relative gene frequencies of cognate tRNAs (RGFCt) to quantify the non-uniform availability of cognate tRNAs in each family of synonymous codons. Using these two quantities, we define two general groups of codons: non-frustrated codons, whose usage in the coding sequences is in proportion to the expected cognate tRNA levels, and frustrated codons, which do not satisfy this proportionality. With this decoding for every codon, we defined the Codon Frustration Index (CFI) as the net frustration of a gene, normalized for its length. Notably, we find that CFI correlates very well with other independent measures of CUB and a high content of non-frustrated codons increases both translation efficiency and mRNA stability. Finally, we show that genes with either a high content of frustrated or of non-frustrated codons are differentially enriched in specific functional classes that typically comprise nucleic acid binding proteins, mRNA processing factors, RNA helicase, and in several transcription factors.