Repository-based plasmid design

Abstract

There was an explosion in the amount of commercially available DNA in sequence repositories over the last decade. The number of such plasmids increased from 12,000 to over 300,000 among three of the largest repositories: iGEM, Addgene, and DNASU. A challenge in biodesign remains how to use these and other repository-based sequences effectively, correctly, and seamlessly. This work describes an approach to plasmid design where a plasmid is specified as simply a DNA sequence or list of features. The proposed software then finds the most cost-effective combination of synthetic and PCR-prepared repository fragments to build the plasmid via Gibson assembly®. It finds existing DNA sequences in both user-specified and public DNA databases: iGEM, Addgene, and DNASU. Such a software application is introduced and characterized against all post-2005 iGEM composite parts and all Addgene vectors submitted in 2018 and found to reduce costs by 34% versus a purely synthetic plasmid design approach. The described software will improve current plasmid assembly workflows by shortening design times, improving build quality, and reducing costs.

Fig 1. Rapid growth in DNA repository size.

(A) Total number of DNA deposits in iGEM, Addgene, and DNASU from 2005 to 2018. (B) Total number of deposits between 2010 and 2018 with linear regression forecasts through 2025 using five (green) or ten (blue) years of past deposit totals. There was a 25-fold increase in the total number of sequence deposits available between 2008 and 2018 and a 99% increase between 2013 and 2018. Continued growth like the last five years would result in roughly 500,000 accumulated deposits between iGEM, Addgene, and DNASU by 2025.

Fig 2. Flowchart of REPP’s plasmid design.

(A) Sequence-specified plasmid design with an example of design output with one synthetic fragment and three PCR fragments. (B) Feature-specified plasmid design with an example where iGEM parts are ordered in the orientation of the target plasmid–REPP finds a vector, BBa_S01987, that matches the user’s design specification. Users can specify their target plasmid by its sequence, composite features, or composite fragments (not shown). (C) Flowchart with the macro-microflow of REPP. Unfilled circles are the start of each process and filled circles are the end. Rounded rectangles are tasks and diamonds are conditions. The flowchart highlights the four steps of REPP’s design: parsing the target plasmid, finding fragments to cover the target, traversing the fragments to create a list of assemblies, and filling the Pareto-optimal assemblies.

Fig 3. Plasmid design of iGEM plasmids from preexisting iGEM parts.

REPP’s design of iGEM composite devices with 98% identity, synthesis with IDT, and iGEM itself as the sole fragment source repository. (A) An alternative to REPP’s designs of each plasmid was their assembly with synthetic fragment(s) alone without PCR amplified sub-fragments. If an iGEM part was too long for a single synthetic fragment, it was divided into multiple synthetic sub-fragments. (B) Another alternative build approach was ordering the iGEM part in a “synthetic gene” where the part is delivered in a pre-cloned plasmid. (C) Cost comparison between plasmids designed with synthetic fragment(s) inserted into linearized pSB1A3 and REPP’s plasmid designs for all iGEM plasmids. For the top quartile of the plasmid designs, REPP’s solution was at least 39% less expensive than synthetic fragments alone. (D) Similar cost comparison between REPP’s plasmid designs and synthetic gene costs. REPP’s solution was always the same cost or less expensive, though often included only synthetic fragments. (E) Frequency distributions for the cost savings of REPP’s plasmid designs versus synthetic fragment(s) (blue) and synthetic genes (red). (F) Length of the iGEM part (insert) versus REPP’s cost savings against synthetic fragment-only plasmid design. A Lowess curve, blue, was created with a windows size of 15%. REPP’s savings were greater than 25% for all plasmid designs greater than 2,000bp and averaged around 37% for plasmids with 3,000bp iGEM parts.

Fig 4. Cost sensitivities.

Plasmid assembly costs for the iGEM and Addgene datasets in response to varied synthesis providers and percentage identities. Median and interquartile range of REPP designed plasmid costs using synthesis costs corresponding to IDT, Eurofins, GeneArt, Genewiz, GenScript, and Twist Bioscience for the iGEM (A, C) and Addgene (B, D) datasets. Synthesis costs were from February 2019. Less expensive synthesis costs like Twist Bioscience’s corresponded with less expensive plasmid designs. REPP’s specification file makes synthesis costs adaptable to future price changes. In (A) the dataset is the same as Fig 3: composite iGEM parts inserted into pSB1A3 digested with PstI and a 98% sequence identity requirement. The plasmids in (B, D) were all uploaded to the Addgene repository in 2018. Median and interquartile range of plasmid costs in response to varied percentage identities for the iGEM (C) and Addgene 2018 (D) datasets. Two-tailed Mann-Whitney U test was applied (ns ≥ 0.05, *p < 0.05, ****p < 0.0001).