Pathway Redesign

  • Pablo Carbonell
Part of the Learning Materials in Biosciences book series (LMB)


In a real scenario, metabolic pathway design is performed through an iterative process. Initial prototypes are designed based on the available pathway knowledge and associated models. However, those initial prototypes are generally not optimized because of the lack of precise understanding of every effect coming from our design choices. Therefore, an initial pilot screening is essential as a way of identifying the most promising designs. However, we should expect low titers and poor performance from this initial screening. A second optimization round is at least necessary. Here, we will discuss pathway redesign and optimization techniques that should help in order to bring up titers and the overall performance in the second iteration, allowing in that way the selection of streamlined designs. Two main approaches are possible: model-based and data-driven. The former will be used in order to redesign the chassis and the enzyme sequences, whereas the latter will be employed for experimental redesign. In order to address these challenges a promising approach is the use of machine learning, an efficient model-free approach that can assist us through the entire redesign process.


  1. 1.
    Angermueller, C., Prnamaa, T., Parts, L., Stegle, O.: Deep learning for computational biology. Mol. Syst. Biol. 12(7) (2016)CrossRefGoogle Scholar
  2. 2.
    Bartsch, S., Bornscheuer, U.T.: Mutational analysis of phenylalanine ammonia lyase to improve reactions rates for various substrates. Protein Eng. Des. Sel. 23(12), 929–933 (2010). CrossRefGoogle Scholar
  3. 3.
    Buß, O., Rudat, J., Ochsenreither, K.: FoldX as protein engineering tool: better than random based approaches? Comput. Struct. Biotechnol. J. 16, 25–33 (2018). CrossRefGoogle Scholar
  4. 4.
    Chen, Y., Li, Y., Narayan, R., Subramanian, A., Xie, X.: Gene expression inference with deep learning. Bioinformatics 32(12), 1832–1839 (2016). CrossRefGoogle Scholar
  5. 5.
    Chollet, F.: Deep learning with Python. Manning Publications (2018).
  6. 6.
    Colwell, L.J.: Statistical and machine learning approaches to predicting protein-ligand interactions. Curr. Opin. Struct. Biol. 49, 123–128 (2018). CrossRefGoogle Scholar
  7. 7.
    Holm, L., Laakso, L.M.: Dali server update. Nucl. Acids Res. 44(W1), W351–W355 (2016). CrossRefGoogle Scholar
  8. 8.
    Khersonsky, O., Lipsh, R., Avizemer, Z., Ashani, Y., Goldsmith, M., Leader, H., Dym, O., Rogotner, S., Trudeau, D.L., Prilusky, J., Amengual-Rigo, P., Guallar, V., Tawfik, D.S., Fleishman, S.J.: Automated design of efficient and functionally diverse enzyme repertoires. Mol. Cell (2018). CrossRefGoogle Scholar
  9. 9.
    Li, W., Godzik, A.: Cd-hit: a fast program for clustering and comparing large sets of protein or nucleotide sequences. Bioinformatics 22(13), 1658–1659 (2006)CrossRefGoogle Scholar
  10. 10.
    Li, Y., Wang, S., Umarov, R., Xie, B., Fan, M., Li, L., Gao, X.: DEEPre: sequence-based enzyme EC number prediction by deep learning. Bioinformatics (2017). CrossRefGoogle Scholar
  11. 11.
    Mura, C., Draizen, E.J., Bourne, P.E.: Structural biology meets data science: does anything change? Curr. Opin. Struct. Biol. 52, 95–102 (2018). CrossRefGoogle Scholar
  12. 12.
    Ortiz, A.R., Strauss, C.E., Olmea, O.: MAMMOTH (matching molecular models obtained from theory): an automated method for model comparison. Protein Sci. 11(11), 2606–2621 (2002). CrossRefGoogle Scholar
  13. 13.
    Richter, F., Leaver-Fay, A., Khare, S.D., Bjelic, S., Baker, D.: De novo enzyme design using Rosetta3. PLoS One 6(5), 1–12 (2011). CrossRefGoogle Scholar
  14. 14.
    Samish, I.: The framework of computational protein design. Methods Mol. Biol. (Clifton, N.J.) 1529, 3–19 (2017). Springer. Google Scholar
  15. 15.
    Schumacker, R.E.: Learning statistics using R. Sage Publications, Thousand Oaks (2014)Google Scholar
  16. 16.
    Segler, M.H.S., Preuss, M., Waller, M.P.: Planning chemical syntheses with deep neural networks and symbolic AI. Nature 555(7698), 604–610 (2018). CrossRefGoogle Scholar
  17. 17.
    Wijma, H.J., Fürst, M.J.L.J., Janssen, D.B.: A computational library design protocol for rapid improvement of protein stability: FRESCO. Methods Mol. Biol. 1685, 69–85 (2018). Humana Press, New York. Google Scholar

Further Reading

  1. A discussion about how computational protein design is expanding synthetic biology applications:Google Scholar
  2. Gainza-Cirauqui, P., Correia, B.E.: Computational protein design – the next generation tool to expand synthetic biology applications. Curr. Opin. Biotechnol. 52, 145–152 (2018). CrossRefGoogle Scholar
  3. The site provides a good introduction to many topics related to scientific computation using Python. The sections about statistics and machine learning are a useful reference material for the methods discussed in this chapter.
  4. An excellent introduction to deep learning using Python:Google Scholar
  5. Chollet, F.: Deep learning with Python. Manning Publications (2018).Google Scholar

Copyright information

© Springer Nature Switzerland AG 2019

Authors and Affiliations

  • Pablo Carbonell
    • 1
  1. 1.Manchester Institute of BiotechnologyUniversity of ManchesterManchesterUK

Personalised recommendations