• First notebook.

• Steve,s notes from the meeting with Jacob, Bill and me on 9/6/2018 where we discussed the results from the first notebook.

• Methods to interpret the models, 1 and 2

• Johannes method to cluster UniProt towards faster blast (used in ss prediction)

• Spectrum kernel technique. To mke it work with our high dimentionality, THIS might be a solution (found it at stack exchange).

• second notebook

• Steve's notes from meeting with Johannes on 2/10/2018

• A wrap-up including figures presented at the asbmb meeting in Utah, 2018

• Our predictions look good on a very small set of known activators from yeast and one from drosophyla. We have to test the model in a bigger validation set before publication.

• Interestingly, in Stark publication a region at about positions 250 was found to be tAD in Drosophila. The reported gene used for their findings is developmental core promoter, which is inducible (same as the one we are using in our experiments with ARG3!!).¹ We found the same using our predictor!!

• Stark data was used to build a null hypothesis distribution to compute a p-value for correlation of a vector composed of prediction scores with a vector consisiting on Stark's measurement data (rolled or concatenated). Notebook and libraries.

• unfinished1

• unfinished2

• First Summary

Addgene ID 71169, section Cloning of luciferase reporter vectors of this publication.