This presentation is a quick overview of all of my projects for Cubist Pharmaceuticals while I was working as an intern. I have worked on 5 projects for automating discovery technologies lab workflows and all of them are described in full detail in this presentation.
3. BACKGROUND
Myself
Junior studying Computer Science at Wentworth
Institute of Technology
Interested in bioinformatcis, cheminformatics, and
medical informatics
Worked at MAVERIC for almost a year designing scripts
for data mining software
Project
Automating processes that were being done manually
Making programs/workflows user friendly and flexible
Reducing human error and making processes more
efficient
4. BACKGROUND - DATA CHART
Catego Program/Script Description Technologies
ry Used
NMR GLUE Software Integration and automation of C#
fragment screening
NMR GLUE Overlord Integration and automation of C#, Overlord,
fragment screening – overlord Automap, Tecan
workflow
CM Solubilise Method One of the steps in the DMSO Tecan, VBScript
stock solution creation procedure
CM Matrix Rack Detection and calculation of C#, Overlord,
Merger remaining volumes of DMSO Automap, Tecan
stock solutions in barcoded
Matrix rack tubes
CM Project Aliquoter Dispensing and dilution of DMSO C#, Overlord,
liquid samples Automap, Tecan
5. GLUE SOFTWARE - BACKGROUND
(INTEGRATION AND AUTOMATION OF FRAGMENT SCREENING)
fragment screening: a process of mixing various compounds
with an enzyme and recording the compounds that stick to it
(recording the hits).
Since the compounds being mixed with the enzyme are very
small a sensitive method is needed to detect the hits.
One of the most common fragment screening methods
employs NMR . This method is very effective but the downfall
is that it can be slow. One way to go around it is with the
process of pooling (pool size 4-24).
The bigger the pool size the fewer times one has to run the
NMR, however the bigger the pool size the more time one has
to spend deconvoluting the sample made from all the
compounds.
The whole process of fragment screening can approximately
take up to 10 days or more. It includes a lot of manual labor
mainly when preparing the samples.
6. GLUE SOFTWARE
(INTEGRATION AND AUTOMATION OF FRAGMENT SCREENING)
Problem Solution Lines
• Manually create sample tracking • Automatically create sample 7893
sheets, run files tracking sheets, run files C#
• Pool and analyze data in a very limited • Automatically pool and allow
way for data analysis
• Process data in Excel • Process data in an SDF file
Benefit
• Click a button which automatically cuts down accidental user errors. Days of work
becomes minutes of automation
• Uses/manipulates SDF files instead of Excel files which adds a lot more flexibility
• Has a built-in project manager which organizes SDF files, sample tracking sheets,
runs files in a very easily accessible manner
• All of the data analysis, sample tracking files, follow-up files are saved in the specified
folders
7. GLUE SOFTWARE - WORKFLOW
(INTEGRATION AND AUTOMATION OF FRAGMENT SCREENING)
8. GLUE SOFTWARE - USER INTERFACE
(INTEGRATION AND AUTOMATION OF FRAGMENT SCREENING)
9. GLUE OVERLORD
(INTEGRATION AND AUTOMATION OF FRAGMENT SCREENING)
Problem Solution Lines
• Perform pooling manually • Design a set of scripts and a workflow 1909 C# in
based on CSV files within the Overlord to be able to OSC
automate the pooling process
Benefit
• Instead of pipetting manually set up the workflow, run it and then walk away until it is
finished
List of Features
• Barcodes/no barcodes
• Lays out destinations on deck and gets
source plates one at a time
10. SOLUBILISE METHOD
Addition of DMSO to compound powders; one of the steps in the DMSO
stock solution creation procedure.
Problem Solution Lines
• Converting Mosaic output • Designing a script that automatically converts 294 of
file into Tecan worklist a Mosaic output file into a Tecan worklist VBS in
manually GSC
Benefit
• Automatically generates worklist from Mosaic file
• Error check on rack barcodes
• Reduce human error
Gemini Script
Component
(GSC)
11. MATRIX RACK MERGER
Detection and calculation of remaining volumes of DMSO stock
solutions in barcoded Matrix rack tubes.
Problem Solution Lines
• Merge CSV files • Design a set of scripts that update fields 258 C#
• Update fields based on a based on a formula in OSC
formula • Detect any errors based on the positioning
• Detect errors manually of the tubes in the matrix rack
• Be able to handle 12 racks
Benefit
• MRMerger Method can handle processing up to 12 racks on the Automap in one run.
• Allows the user to walk away during processing
• Adds efficiency and removes user error by doing the file processing automatically.
Overlord Script Component (OSC)
12. PROJECT ALIQUOTER
Dispensing and dilution of DMSO liquid samples.
Problem Solution Lines
• Each plate had to be • Design a set of scripts and multiple 9875 C# in
dispensed/diluted manually workflows within to automate the OSC
or by operating Tecan process for multiple plates.
separately • Defrost
• Each plate is run • Cap/de-cap/seal
separately • Scan 2D barcodes, scan 1D barcodes
• Check/handle errors
• Show virtual graphs of plate maps
• Make it flexible so it handles all kinds
of plates
Benefit
• Biggest advantage is that one could actually walk away from the run after setting
everything up which only takes a couple of minutes
• What could have been done manually in 2 hours could be achieved automatically in
30 minutes (person does not need to be there only for setup)
13. PROJECT ALIQUOTER WORKFLOW - 3000 FEET VIEW
Inputs
Validate Source Plates
Validate Dest Plates
Choose Method
Solution Case 1 Solution Case 2
14. PROJECT ALIQUOTER – 30 FEET VIEW
Required Inputs Validate Destination Plates
Validate Source Plates Decide on Solution
Solution Case 1 Solution Case 2
15. PROJECT ALIQUOTER – 30 FEET VIEW
Inputs
• Browsing for a Mosaic control file
• Deciding/Selecting if defrosting is
needed
• Loading required plates onto hotels
16. PROJECT ALIQUOTER – 30 FEET VIEW
Validate
Source
Plates
• Validating source barcodes against
barcodes listed in Mosaic control file
• Recording positions of source plates
17. PROJECT ALIQUOTER – 30 FEET VIEW
Validate
Destination
Plates
• If defrosting is selected then validate
destination plate barcodes and defrost
source plates simultaneously
• If defrosting is not selected then only
validate destination plate barcodes
18. PROJECT ALIQUOTER – 30 FEET VIEW
Required Inputs Validate Destination Plates
Validate Source Plates Decide on Solution
Solution Case 1 Solution Case 2
19. PROJECT ALIQUOTER – 30 FEET VIEW
Choose Method/Decide on Solution
• If source plates do not need to be
capped/decapped go to Case 1
• If source plates need to be
capped/decapped go to Case 2
• Flexible in a way so that if more
solutions are needed one can easily
do so
20. PROJECT ALIQUOTER – 30 FEET VIEW
• Puts all source plates on Tecan deck
(sources do not need capping/decapping)
• Iterates through destination plates and
process them one by one
Solution Case 1
21. PROJECT ALIQUOTER – 30 FEET VIEW
Required Inputs Validate Destination Plates
Validate Source Plates Decide on Solution
Solution Case 1 Solution Case 2
22. PROJECT ALIQUOTER – 30 FEET VIEW
• Puts one source plate and one destination
plate on Tecan deck (source plate needs
capping/decapping)
• Processes them individually
Solution Case 2
24. CHALLENGES
Getting used to new programming environment and
its rules
Adapting scripts to changes
Adapting/learning how to communicate with
scientists
Learned about the limitations of hardware
Making workflow/scripts flexible
25. CHALLENGES (CONTINUE)
Making workflow/scripts flexible
Problem
Code is nothing but a translation of an algorithm. With any
outside change algorithm changes as well
Update codes to new hardware
Average computer users can not change the code so one
always needs a programmer around
Solution
Use external files which reflect the outside change then feed
the files into the algorithm/code
Any person can change the file because no programming
knowledge is required
Files reflect the change and algorithm will recognize the
change from the files
26. CHALLENGES (CONTINUE)
Making workflow/scripts flexible
Example
One of the examples is the predefined labware type file which
includes all of the labware type + information about them
When a new labware is added there is no need for a
programmer to change the code – file reflects the change and
is fed into the algorithm
27. LESSONS
Schools do not teach C# so I am glad I got to write
almost 20000 lines of code in MV C# 2008 Express
Working with million dollar equipment raised my
confidence
Learned about Overlord and Gemini software
environments which are widely used
Never before have I tackled a project on which I
worked for over 3 months in real world
Making code flexible so that changes do not require
a lot of effort