Unified rational protein engineering with sequence-based deep - - PowerPoint PPT Presentation

Unified rational protein engineering with sequence-based deep representation learning

Ethan C. Alley1,2,6, Grigory Khimulya6,7, Surojit Biswas1,3,6, Mohammed AlQuraishi 4 and George M. Church 1,5*

1

1Wyss Institute for Biologically Inspired Engineering, Harvard University, Boston, MA, USA.

2MIT Media Laboratory, Cambridge, MA, USA. 3Department of Biomedical Informatics, Harvard Medical School, Boston, MA, USA. 4Department of Systems Biology, Harvard Medical School, Boston, MA, USA. 5Department of Genetics, Harvard Medical School, Boston, MA, USA. 6These authors contributed equally: Ethan C. Alley, Grigory Khimulya, Surojit Biswas. 7Unaffiliated: Grigory Khimulya.

2

Evolutionary Approaches Structural Approaches

3

Recurrent Neural Network

http://colah.github.io/posts/2015-08-Understanding-LSTMs/ 4

Long Short Term Memory (LSTM)

5

multiplicativeLSTM

6

multiplicative LSTM (mLSTM)

7

8

9

10

11

Conclusions

• UniRep features embody a subset of known protein characteristics
• Capable of determining the stability and functionality of proteins
• UniRep is able to generalize to unseen portions of the protein fitness

landscape

12

Future Work

• Extend UniRep to generate novel proteins
• Use UniRep to vectorize protein to determine evolutionary information

13