Lab Notebook

Expression Prediction and RBS optimization


Week 1
  • Recruitment of the team completed
  • Review of the original work by Salis et al. for better technical understanding of the problem statement
  • Review of other work on predictive design of ribosome binding sites - thermodynamic models, kinetic models, stochastic models, machine learning models
Week 2
  • Review of other work on predictive design of ribosome binding sites - thermodynamic models, kinetic models, stochastic models, machine learning models
  • Installing the ViennaRNA library for RNA sequence analysis using Python
  • Downloading a dataset of 1014 RBS sequences with experimentally determined expression levels from Reis and Salis (2020)
Week 3
  • Defining the interaction between the rRNA and mRNA so as to have minimum hybridization energy
  • Including the effect of spacing between the SD sequence and the start codon
  • Defining the interaction between the rRNA and the ribosomal S1 protein
  • Accounting for the transient regional folding of the mRNA that can hinder RBS accessibility
Week 4
  • Including the presence of standby sites upstream of the SD sequence to which the ribosome can bind if the SD sequence is transiently folded
  • Accounting for the effect of various start codons on translation efficiency by using data from Hecht et al. (2017)
  • Training and testing different machine learning models to see which works best for the data at hand


Week 1
  • Downloading a dataset of 16779 RBS sequences characterised by FlowSeq from Reis and Salis (2020) and testing the model on it
  • Reading up on different optimization algorithms that can be used to "mutate" an RBS sequence in silico in order to achieve a desired expression level
Week 2
  • Simulated annealing and gradient descent optimisers were created and tested
Week 3
  • Optimiser performance was tested by varying its intrinsic parameters, work is ongoing
  • A partially working genetic algorithm was created, work is yet to be finished on this
  • The two working optimisers were used to generate RBS sequences with varying expression levels. These sequences will be used in the wet lab using GFP-expressing Lactococcus lactis to see if the optimization was successful.
Week 4
  • All India iGEM Meet


Week 1
  • Genetic algorithm completed. Found to have the best performance.
Week 2
  • Attempted to combine different aspects to the Optimization algorithms.
  • Started work on adding new parameters to the model to improve accuracy of predictions. Literature review on RNA folding kinetics and RNA degradation was initiated.
Week 3
  • Literature review continued
Week 4
  • Starting working with the Kinfold library of ViennaRNA to find the folding time. There was a need to integrate command line operations with Python code.


Week 1
  • Kinfold used to generate folding time data for the dataset.
  • Prototype for degradation model built
Week 2
  • Folding kinetics and degradation did not improve model performance.
  • Model finalised without using these parameters.

Probiotic Treatment For Homocystonuria


Week 1
  • Recruitment of the team completed
  • Installed and initialized MATLAB and SimBiology. Understood how to use SimBiology for solving ODEs and running mass kinetic models
Week 2
  • Initial literature review needed before starting off with the modelling
Week 3
  • Identified the genes that must be inserted in L. lactis to help it successfully convert cysteine to methionine and identified which gene must be knocked out.
  • Initial transcription modelling process started off by modelling the binding of the RNA polymerase to the transcription factor
Week 4
  • Completed transcription modelling by modelling the rate of transcription by the RNA polymerase.
  • Started off with the translation modelling by modelling the binding of the Ribosome to the RBS of the mRNA


Week 1
  • Completed translation modelling by modelling the rate of translation by the ribosome binding site.
Week 2
  • Literature review for what kill switches are ideal for the project
Week 3
  • Made a list of all possible kill switch designs that can be used. Identified that a glucose concentration based kill switch is ideal in this case.
  • Started off with identifying the genetic circuit for the kill switch
Week 4
  • All India iGEM Meet


Week 1
  • Identified the most efficient promoter which is induced by the glucose concentration difference between stomach an small intestine.
Week 2
  • With the kill switch circuit confirmed, we started off modelling the circuit on SimBiology
Week 3
  • Modelled the conversion of homocysteine to methionine by the engineered enzyme in L. lactis
Week 4
  • Compiled all the small processes modelled to obtain the final model for the system


18 Aug

  • Randomized primers for both UTR regions received.
  • Lyophilized primers made to 100uM conc and resuspended. Dilutions of 10uM are made.

Sequences are referred to by serial numbers
Optimization Algorithms used to generate the sequence:
SA - Simulated Annealing
GD - Gradient Descent
GA - Genetic Algorithm

H, L = High, Low Expressing

19 Aug

  • Colony PCR (Attempt 1) : Template solution was made by picking colonies and dispersing in aq medium.

21 Aug

  • Gel Check : All PCR products ran on agarose gel; 6, 8, 13, 16 amplified

23 Aug

  • Colony PCR (Attempt 2) (unamplified sequences) : Reduced annealing temperature to 54, 53
  • Gel Check :
    • 9, 10, 15: Strong bands
    • 7, 11: weak bands

24 Aug

  • PCR Attempt 3 (All Samples) : 1uL pure plasmid taken as template

25 Aug

  • Gel Check : 4, 5, 7, 8, 9 14 Not amplified

27 Aug

  • PCR Attempt 4 (Unamplified samples) : Reduced annealing temperature to 53, 52
  • Gel Check : only 9 Amplified

28 Aug

  • PCR Attempt 5 (Unamplified samples) : Touchdown PCR 56-52C
  • Gel Check : 4, 5 Unamplified

30 Aug

  • PCR Attempt 6 (Unamplified samples) : 50uL reaction separated into 2 tubes of 25uL
  • Gel Check : both fragments (4, 5) amplified

  • All fragments have been amplified!

31 Aug

  • 0.4uL Dpn1 added to all samples
  • Gel Extraction : (Qiagen Kit protocol)
    • Plasmid conc :
      • 3,4,5,6,7,8,9,10 - around 10ng/uL.
      • 11,12,13,14,15,16 - around 30-40 ng/uL
      • 5 - 2ng/ul


2 Sep

  • Transformation + Plating
    • Electroporation was done for all the samples in L. lactis competent cells.
    • Samples were incubated for 1hr after which they were spread plated and incubated for 24 hours.

3 Sep

  • Colonies in all plates except samples 4 and 14. 10 colonies were observed in plate 3.
  • Colonies were picked and cultured overnight. (2 cultures from each non-randomized sample, all 10 cultures of each colony from the randomized sample 3.)

4 Sep

  • Glycerol added and strains stored in -80C

5 Sep

  • Transformation and plating (Attempt 2) (Samples 3,4,14) (Working to obtain more strains of randomized sample 3)

6 Sep

  • No colonies observed

8 Sep

  • Transformation and plating (Attempt 4) (Samples 3,4,14)

9 Sep

  • No colonies observed

11 Sep

  • Transformation and plating (Attempt 5) (Samples 3,4,14) (Slightly reduced antibiotic concentration)

12 Sep

  • Golden Gate Assembly set up from left over plasmid for 3,4,14

14 Sep

  • Transformation and plating (Attempt 6) (Samples 3,4,14)

15 Sep

  • 1 colony for sample 3. None for 4 and 14.

17 Sep

  • Transformation and plating (Attempt 7) (Samples 3,4,14)

18 Sep

  • 2 colonies in sample 3. None for 4 and 14.

21 Sep

  • All strains (13 variants from randomized sample 3, 11 model generated sequences) streaked on agar plates

22 Sep

  • Colonies picked and Overnight 1ml cultures made and incubated for 14 hrs

23 Sep

  • Fluorescence Assay
  • First batch of results obtained.

25 Sep

  • Colonies picked from streaked plates and cultured overnight.

26 Sep

  • Plasmid isolation (Qiagen Kit protocol)
  • Plasmid concentration obtained were very poor.


2 Oct

  • All strained streaked from glycerol stocks.

3 Oct

  • Colonies were picked from streaked plates and cultured overnight.

4 Oct

  • Plasmid isolation (Qiagen Kit protocol).

5 Oct

  • Samples sent for sequencing.