We created a software tool that uses a DNA strand input, and defines a list of crRNA sites output. These crRNA will bind to the DNA with a strength previously specified by the user, as a percentage of the fold change of the dCas9. It also calculates the approximate structure of the crRNA and displays first the ones which disturb the dCas9 handle the least. To achieve this, it incorporates open source code from RNAfold [1] and code used in the research of Iman Farasat and Howard M. Salis [2] generously provided by them. The aim is to use this tool in order to design PAM proximal promoter sequences, as there is currently no analytical method available for prediction of binding strength between a dCas9 enzyme and PAM sequence.

The fold change is the fold difference in gene expression between the case when the dCas9 is bound to the promoter and when it's not bound. Since binding of dCas9 in a CRISPRi system prevents transcription by blocking access to the promoter, the strength of binding of dCas9 determines the fold-repression.

To estimate the relation between these, data from two papers is used. For a given DNA sequence one paper presents the fold change in binding of dCas9 for different number of mutations [3]. The second paper presents the repression activity of a single mutation at different locations[4].

In both cases the RNA:DNA binding is calculated with the Cas9Calculator and then plotted on a logarithmic scale against the fold change (Fig. 1) or one over the repression activity, respectively (Fig. 2). With the exception of few points, correlations seems to be linear and hence the relation is fitted with the following formula:𝑃=𝑒^(−𝛽(𝑚𝑎𝑥𝐺−𝐺))

Where maxG is the Gibbson free energy when there are no mismatches and G is the energy for the given state. P is the fold change between the two states. The coefficient %beta = 1.003074571 is the average of all individual coefficient satisfying the formula (with the exception of the few points noted above).

Our aim is to launch a website which will run the code described above in real time making it easier to use, thus reaching a wider range of users. So far an interface has been created and the code was successfully run on a local server (Fig. 3,4,5). The tool could be further improved by increasing its precision or decreasing the time it takes to return results. This could happen by fitting the data from section 2 with a different function or developing a more efficient algorithm for creating crRNA sequences. However, the idea behind this tool is simple but could still prove to be efective in the engineering of CRISPR/Cas9 systems.