Data Encoding

Encoding Method:
Currently, we support six encoding methods: (i) MaxDensity (exhibits the highest density of information - two bits in every nucleotide) and (ii) NoHomopolymer (designed to prevent homopolymers in the DNA strand) are based on a de-novo synthesis approach. (iii) LinearChain (based on building DNA strands by hybridising and ligating oligos in a linear order, see [2]), and (iv) LinearBinomial (based on the same assembly strategy but encodes data in a combinatorial manner, see[1,2]). (v) PolyChain (based on building DNA strands by hybridising and ligating oligos in a positional order, see [2]), and (vi) PolyBinomial (based on the same assembly strategy but encodes data in a combinatorial manner, see[1,2]).

References

Error Correction

Outer error Correction:
Outer error correction adds redundant information to every codeword to enable the correction of errors in the data, like substitutions, insertions, and deletions, see [1]. Currently, only the Reed Solomon code is supported but the percentage of redundancy can be adjusted by the user.

Inner error Correction:
The inner error correction scheme is applied to the data after the outer error correction scheme is applied and adds redundant codewords to make it more resilient against the loss of entire codewords. Currently, we support the LT code [2] only, where the user can specify the percentage of additional codewords. Block-by-block assembly based DNA encoding methods have an intrinsic error correction, since sequencing reads can be aligned to the oligo library used for assembly. This allows to recover lost data as long as enough correct oligos are present in the sequencing reads. Still, in most cases it is beneficial to add (at least) inner error correction.

References

Synthesis Simulation

The simulation step allows to simulate errors for two different kinds of processes : (i) nucleotide by nucleotide synthesis (e.g. by phosphoramidite or enzymatic chemistry), or (ii) assembly of shortmers. Naturally these approaches exhibit completely different errors and are, thus, simulated in different manners.
Synthesis
The simulation of DNA synthesis errors follows the approach by Schwarz et Al. [1] and is based on published error information for different combinations of synthesis methods and error correction methods [2]. The available methods are CSO (column synthesized oligos) and MBOP (microarray-based oligo pools). Each method is characterised by a base line error rate and a distribution of error rates for insertions, deletions, and mutations.
Assembly
In the assembly step, the oligonucleotides are mixed in the order specified by the pooling algorithm. Our pooling method attempts to simulate realistic hybridisation / ligation results by (i) drawing random quantities of oligos, (ii) randomly selecting pairs of oligos, (iii) calculating the most likely conformations using a nearest neighbor model, and (iv) randomly drawing from these. The result of this simulation includes oligos of different lengths and potentially incorrectly hybridised oligos.
The user can specify the mean and standard deviation of the oligo quantities to be drawn from a normal distribution.

Storage Simulation

In this step, users can simulate the storage of the DNA sequences that encode your data. Three different storage conditions can be selected: (i) room temperature, (ii) permafrost, and (iii) Biogene capsule. The simulation of DNA storage in vitro follows the most widely used kinetic model for DNA decay (caused by strand breaks), which assumes a constant degradation rate (over time and over nucleotides). Currently, three rates are available: (i) storage at room temperature (5\*10^-3/nt/year), (ii) storage in permafrost (5.5\*10^-6/nt/year [1]), and (iii) storage in an anhydrous nd anoxic atmosphere inside a hermetic capsule develop by Imagene for long-term conservation of biospecimen at room temperatures (1\*10^-7/nt/year) [2].

References

Sequencing Simulation

In this step, users can simulate the read-out (sequencing) of the DNA sequences. There are three different sequencing technologies to chose from: (i) Oxford Nanopore, (ii) Illumina, and (iii) PacBio. The DNA sequencing process follows the approach followed by Schwarz et Al. [1] and includes six sequencing technologies: (i) single-read and (ii) paired-end by Illumina (Schirmer et al., 2016) [2], (iii) subread and (iv) CCS methods by PacBio (Weirather et al., 2017) and (v) 1D and (vi) 2D sequencing by Oxford Nanopore (Weirather et al., 2017)[3]. Every method is characterised by a raw error rate as well as error rates and probabilites for three errors: insertions, deletions, and mutations. The number of errors for a chosen method is calculated by multiplying the sequence length with the raw error rate of the method. The selected type of each error is based on the weights of each error class, positional weights or restrictions and a random number generator. The errors are applied sequentially, allowing for error cross-talk. For example, if a deletion led to the formation of a triplet with a high error probability, this triplet will be included in urther error evaluations.

References