Your browser doesn't support the features required by impress.js, so you are presented with a simplified version of this presentation.

For the best experience please use the latest Chrome, Safari or Firefox browser.

machine learning with LARA,
SiLA2/AnIML providing experimental data for ML/AI and cheminformatics

mark doerr, patrick courtney, uwe bornscheuer and stefan born

institute for biochemistry, university greifswald/TU-berlin/SiLA
2019.07.09

uni-logo
Any element with the class="notes" will not be displayed. This can be used for speaker notes. In fact, the impressConsole plugin will show it in the speaker console! Press ctrl-C to activate the console
greifswald-map
* lara intro

biocatalysis - e.g. transamination reaction

- highly selective enzymes replace conventional chemistry

TA

Calvelage, S.; Dörr, M.; Höhne, M.; Bornscheuer, U. T., Advanced Synthesis & Catalysis 2017, 359 (23)

* lara intro

the greifswald protein screening platform LARA

lara
* lara intro

LARA - real robotics at the institute for biochemistry

LARA movie

* protein screening engineering * findind the right enzyme in 1E5 to 1E9 variants * lara movie

LARA - components top view

lara top
* protein screening engineering * findind the right enzyme in 1E5 to 1E9 variants * lara movie

scientific data - structure

scientific-data
* In the very early days of personal computing, I was wondering, why the computer was not used

scientific work and data

how do we organise / structure scientific experiments that they are reproducible, even in 100 years ?

how do we store the scientific data, even to be read in 100 years ?

how do we organise the scientific data, that new knowledge can re-interpret / re-evaluate the old data ? (coping with evolution in science)

* In the very early days of personal computing, I was wondering, why the computer was not used in every scientist should have a laptop to document his experiments share data , faster evaluation, search, combining results, re-interpret when more knowledge First devices every single devices has its very own software HPLC, NMR own data formats only propieatary sotware to open the formats later robotics very limited view only control so I decided to create my onw software with the massive help and back of the open source community everything is out there....

holistic approach of the LARA suite

LARA-workflow
* In the very early days of personal computing, I was wondering, why the computer was not used

LARA suite - architecture

LARA-structure
* In the very early days of personal computing, I was wondering, why the computer was not used

LARA suite - main components

  • robot project planning
  • robot process design / code generation
  • process control / scheduling (planned)
  • data collection
  • data evalution and visualisation
  • people / experimentalists
  • compound and device database
* feature

LARA suite - techniques

what protein engineering questions can be addressed by machine learning ?

* In the very early days of personal computing, I was wondering, why the computer was not used

example ML session

finding better variants of the ProteinaseK enzyme (Liao et al, Biotechnology 2007)

  1. start with an initial sets of enzyme variants
  2. create a math. model of the sequence-activity relationship
  3. e.g., simple linear model and Lasso
    4. scikit
  4. finding optimal regularisation parameters
  5. learn from 500 different subsets
  6. apply the learned coefficients
  7. provid suggestions for the next round of experimentation
  8. learn from new variants
  9. restart from step 6

Liao et al, Biotechnology 2007

* In the very early days of personal computing, I was wondering, why the computer was not used

LARA machine learning demo

* LARA

LARA - open source code repositories

gitlab.com/LARAsuite

github.com/LARAsuite

we need you as a open source developer / tester !

what is sila_logo ?

sila_logo

sila-standard.org

* feature

who is sila_logo ?

* feature

sila_logo 2 lab automation communication standard

sila2 structure
* SiLA 2

sila_logo 2 - applications

sila2_integration
*

SiLA2 - repositories

gitlab.com/SiLA2

raspi Raspberry Pi repository: sila_python/raspberry_pi

sila-standard.org

image source: https://www.raspberrypi.org

summary

  • the power of open source ('matters') - construction of very powerful tools with standard solutions
  • fast, script based / automatized data processing / machine learning with LARA, using one single language paradigm to access all data
  • SiLA2 based communication
  • highly structured, high-quality data from one uniform data source
  • very easy access to data in LARA database (no SQL required)
Software that makes the usage of everything less complex

acknowledgements

robot hard- and software

  • Stefan Born (TU Berlin)
  • Peter Neubauer with his group (TU Berlin)
    • Sebastian Hans (TU Berlin)
    • Shaon Debnath (TU Berlin)
  • Johannes Kabisch with his group and associates (TU Darmstadt)

SiLA team

  • Daniel Juchli (wega-it.com)
  • Maximilian Schulz (unitelabs.ch)
  • Stefan Koch (equicon.de)
  • Oliver Peter (idorsia.com)
  • Patrick Courtney (tec-connection.com)

general

  • uwe Uwe Bornscheuer (University Greifswald)

THANX !