5. Title: Yearly Progress Meeting (Starting my 3rd PhD year)

5.1. Date

02062020

5.2. Objective

• To reflect on agreements from the Go-No GO meeting in 2019, whether they have been followed or changed and why.

• Make a plan for coming years.

5.3. Method

• One to one meeting with my supervisor

5.4. Next Steps

• Convert all my code slowly to the format of the Jupyter Book in order to have a better visualization and story telling of the code.

• Connect always data (DOI in a data repo like 4TU) in the code (in python, refers to a link in 4TU maybe to download the files) and story line using jupyter notebooks. This is a structure to foster reproducibility of the data analysis.

• For the reproducibility of experiments keep using this Experimental Journal :)

5.4.1. Code Related

• Idea to predict changes in genetic interactions from data from essential genes from SATAY. For every mutant there are genetic interaction data and the idea is to couple this with the essential genes to investigate how the essential genes in a specific background tell us about the difference in genetic interactions.

  • For example the samples are every tested mutant and the features could be the density of transposon per gene in the genome and the type of interaction is registered between the knockout gene and every gene of interest.

• Do more testing on the ML workflow that from protein domains is capable of predicting synthetic lethality in yeast. Specifically find :

  • The most important features (protein domains that are specific to SL)

  • Change the negative dataset (nonSL), make it bigger or change it, to test the specificity of the model.

• Use this idea of features using protein domains to try to predict Bem1 rescue partners .

• Clean the code for the common Go-terms and common interactors partners , by excluding non informative Go-terms , that do not express nothing.

# remove the meaningless GO terms
    bad_GO = ['cellular_component','biological_process', 'molecular_function', 'not_yet_annotated','other']