Predicting the Genetic Stability of Engineered DNA Sequences with the EFM Calculator

Abstract

Unwanted evolution can rapidly degrade the performance of genetically engineered circuits and metabolic pathways installed in living organisms. We created the Evolutionary Failure Mode (EFM) Calculator to computationally detect common sources of genetic instability in an input DNA sequence. It predicts two types of mutational hotspots: deletions mediated by homologous recombination and indels caused by replication slippage on simple sequence repeats. We tested the performance of our algorithm on genetic circuits that were previously redesigned for greater evolutionary reliability and analyzed the stability of sequences in the iGEM Registry of Standard Biological Parts. More than half of the parts in the Registry are predicted to experience >100-fold elevated mutation rates due to the inclusion of unstable sequence configurations. We anticipate that the EFM Calculator will be a useful negative design tool for avoiding volatile DNA encodings, thereby increasing the evolutionary lifetimes of synthetic biology devices.

Keywords: computer-aided design (CAD); design-build-test cycle; genetic engineering; genetic robustness; hypermutable site; metabolic engineering.

Figure 1.

(a) The EFM Calculator accepts a DNA sequence as input, predicts two types of hypermutable sites in this sequence, and summarizes the overall prediction of instability with a RIP score. It outputs an HTML file of hypermutable sites on an interactive drawing of the sequence and in a table. The RIP score represents the factor by which redesigning the input sequence to eliminate predicted SSR and RMD mutational hotspots (leaving only the baseline rate of BPS mutations) could theoretically reduce the overall rate of mutations that contribute to the evolutionary failure modes of this DNA sequence when it is deployed in a bacterial host. (b) The original version of a genetic circuit that expresses GFP (BioBrick part T9002) receives a RIP score of 486.3. A redesigned version (T9002-E), shown experimentally to have a longer evolutionary half-life by Sleight et al., receives a more stable RIP score of 101.6.

Figure 2.

(a) Distribution of RIP scores predicted by the EFM Calculator for parts with lengths of >50 base pairs in the iGEM Registry. The colored shading of bars demarcates ten-fold increases in predicted mutational instability. (b) EFM Calculator output for BioBrick part E1010, one of the top ten most used coding sequences in the Registry. The RIP score of 178.3 indicates that the mutation rate in this part is predicted to be 178 times the baseline rate of base-pair substitutions. Three simple sequence repeats contribute to this genetic instability prediction.