Expanding the Compound Management toolkit: leveraging technology to streamline sample QA/QC in high throughput environments

Expanding the Compound Management toolkit

Pierre BaillargeonHTS/Lead Identification

About Scripps Florida

The Scripps Research Institute © 2012 – All rights reserved

I. Introduction to Lead Identification and Compound Management at Scripps Florida

II. Review of issues with DMSO solvated compounds in HTS libraries and methods for addressing these issues

III. An introduction to the Plate Auditor

IV. Impact of Plate Auditor results on our CM & HTS operations

Integrated Assay Development, uHTS and Compound Management

• >982K Compound Screening Collection• >622K Proprietary (largest in

academia, ~30k unique compounds, focused sub-libraries, professionally curated)

• >360K Public Domain (NIH)• Integrated Compound QC (LC-MS, Plate

Auditor)

The Lead Identification Laboratories at Scripps


Where robotics, chemistry and biology join forces to help discover new drugs

• Located in Jupiter, FL• >175 targets screened in over 185 primary

campaigns to generate more than 60 million data points for academic & industrial collaborators

• S1P1 drug candidate entered phase 1 clinical trials in January 2011

Issues With Compound Libraries





IV. Practical applications of Plate Auditor

Compound Management Challenges


Job cuts

Budget cuts limiting technology investment

Building a library to meet new research requirements

Sample data management

Integration of automation technology

New automation technology

Sample quality

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

16.70%

16.70%

66.70%

66.70%

33.30%

33.30%

66.70%

What are your primary concerns in your current role or your department?

Graph reproduced from the 2012 IQPC ‘Setting up Compound Management Globally’ Compound Management report

http://www.compoundmanager.com/uploadedFiles/EventRedesign/UK/2012/May/11314006/Assets/IQPC---Compound-management-060112.pdf

% of respondents

http://www.compoundmanager.com/uploadedFiles/EventRedesign/UK/2012/May/11314006/Assets/IQPC---Compound-management-060112.pdf

Compound Management Responsibilities & Customers


Compound Management

HTS

DMPK & Pharm

External

Collaborator

Therapeutic Areas

Compound

Vendors

Medicinal Chemistry

What role does CM play at Scripps?• HTS Library Procurement

• HTS Library Stewardship

• HTS Library QC (via LC-MS)

• Cherry pick for HTS Core

• Compound re-synthesis & restock

• Receive & reformat samples from external collaborators

• Ship samples to external collaborators

• Medicinal Chemistry support



Compound Management

HTS

DMPK & Pharm

External

Collaborator

Therapeutic Areas

Compound

Vendors

Medicinal Chemistry

What operational challenges are encountered during these interactions?• Verifying quality of external

samples & sample data

• Degradation or precipitation of samples over time

• Enforcing internal QA/QC standards

• Tracking sample properties & genealogy in a way that can be easily audited in the future

• Managing large sample libraries which are constantly in flux (multiple copies, samples becoming depleted, new acquisitions, etc.)



How do we currently address these issues and what are the limitations?

• LC-MS (time consuming)

• Acoustic auditing (time consuming, labware dependent)

• Manual visual inspection (time consuming, subjective)

• Suspension of solids or precipitated samples (sonicating, heating, solvent change, etc.)



• Degradation or precipitation of samples over time




Opportunities For Error in Sample Distribution


COMPOUND SUBMISSION• Has the correct sample been put into the labware?• Is the volume and concentration of the sample accurately recorded?

COMPOUND PROCESSING• Has the sample store delivered the correct samples?• Has the liquid handling automation pipetted the samples properly?• Has the plate been registered correctly?• Is the sample itself of high quality?

COMPOUND DELIVERY• Do we have a record of what we are delivering?

Room for improvement! How can we QC samples more efficiently?

Introduction to Plate Auditor






Are empty/partially filled wells in Quadrant 1?

Do these wells contain precipitate?

Are these wells partially filled?

My db says this well is

empty. Does it actually

have sample in it?

My HTS detection format is sensitive to purple compounds. Where are they?


Why Do We Need Plate Auditor?

Plate Auditor Applications & Features

• Instant detection of errors from compound reformatting • Detect missed/partially filled wells, quadrant effects

• “Pre-screen” compound plates for HTS assay interferents• Detect colored compounds, suspended materials

• Identify database entry errors in corporate LIMS db• Determine which full/empty wells in plate differ from the full/empty well assignments in corporate db

• Periodic check-ups on the “health” of stored compounds• Detect ppt. formation, evaporation

• Detection of compound solubility• Identify HTS library samples that have “crashed out” of solution


What is Plate Auditor?

• A custom HTS plate reader, incorporating recent advances in machine vision, image analysis, & the spectroscopy sciences

• It saves labor by automatically identifying & annotating issues specific to compound libraries:

• Colored compounds• Precipitate / Crystallization• Low volume / Full wells

• It can be used in “stand-alone” mode, or integrated with automation

• Works with microtiter plates

• Fast: reads a 384 well plate in <1 minute

• All measurements are non-contact

• User-friendly, intuitive analysis software


Plate Auditor Features: Instrument Design


Plate Auditor Features: Color Recognition & Classification

Color Detected Dye Used

Limit of Detection

(uM)

Red Allura Red 35.1

Green Brilliant Green 9.0

Blue Euroglaucine 0.6

Yellow Tartrazine 12.7

Violet Gentian violet 1.3Befo

re P

rocessin

gA

fter

Pro

cessin

g


Plate Auditor Features: Precipitate Detection

Compound plates contain precipitates which are masked by color:

Plate Auditor precipitate analysis is:• Insensitive to color• Able to detect ppt. beyond “naked eye” inspection


Plate Auditor Features: Low Volume Detection

Plate Auditor distinguishes normal wells (“full”) from empty/partially filled wells (“low volume”)

40uL 20uL 10uL 8uL 4uL 2uL 1uL

What can cause a well to have a low volume?• Incorrectly reported

volume in LIMS

• Oversampled by cherry picking operations

• Evaporation

• Liquid handling errors (clogged tips, samples transferred to incorrect wells/quadrants, etc.)


Plate Auditor Features: LIMS/Corporate Database Auditing

RESULTS OF COMPOUND PLATE QUERY IN CORPORATE DATABASE:• Scripps database displays several wells as

empty (white rectangles in figure on left)

RESULTS OF THE SAME COMPOUND PLATE QUERY IN Plate Auditor DATABASE:• Plate Auditor compares instrument results

to records archived in corporate db

• Plate Auditor identifies discrepancies with corporate db records(red “x’s” in figure on left):

• Colored compounds (upper left)• Full wells that corporate db assigns as “empty”


PLATE AUDITOR AUDITS CORPORATE DB RECORDS

Introduction to Plate Auditor






Addressing Opportunities For Error in Sample Distribution


What new information do we obtain with Plate Auditor in the process?• We know which wells have been filled and which have low volume• We know which wells contain colored samples• We know which wells contain compounds that have precipitated

What do we do with this information?• Compare filled & low volume wells against LIMS to verify proper liquid handling & plate

registration• Compare ‘color fingerprint’ of samples against database of known samples to identify

issues with incorrect concentration & samples• Take measures to put precipitated samples back into solution (sonication, heating, etc.)• Keep a historical record of the health of the plate at the point when it was delivered (a

receipt of transaction)

Practical Examples: HTS Library Analysis

This includes Scripps Florida’s internal 622k Drug Discovery library and the NIH’s Molecular Libraries-Small Molecule Repository (MLSMR) 362k library.

To date, Scripps has run over 1,000 plates through Plate Auditor resulting in over 1,000,000 wells which have been analyzed for color, precipitate and low volume artifacts.

How can we leverage all of this new data to improve our workflow?


Practical Examples: HTS Library Analysis


This includes Scripps Florida’s internal 622k Drug Discovery library and the NIH’s Molecular Libraries-Small Molecule Repository (MLSMR) 362k library.

To date, Scripps has run over 1,000 plates through Plate Auditor resulting in over 1,000,000 wells which have been analyzed for color, precipitate and low volume artifacts.How can we leverage all of this new data to improve our

workflow?

2) Use sample color to track fidelity of CM processes & sample quality3) Use sample color as a predictor of sample performance in HTS4) Precipitate results measure effectiveness of solvation SOPs

5) Using precipitate detection to monitor library health

1) Identify LIMS database errors in real time

6) Other interesting observations: Labware QC

Plate Auditor data usage at Scripps Florida


1) Well data classified via Plate Auditor GUI

2) Well data merged with sample information from Corporate Plate DB and stored in Plate Auditor DB

3) Resulting data set exported from Plate Auditor DB and merged with HTS, Cheminformatics and LC-MS results

By making Plate Auditor data available in the same web-based informatics portal as other HTS data, project biologists & chemists can easily examine HTS library samples for volume, color and precipitate issues

… examining the plate detail view reveals a pattern in quadrant 1…

Identifying LIMS database errors in real time

When checking reformatted plates, Plate Auditor finds high # of LIMS disagreement wells…

… checking the LIMS record reveals samples in quadrant 1 are not registered …


Identifying LIMS database errors in real time

… CM team updates LIMS record and runs plate through Plate Auditor to confirm the corrected LIMS record matches the physical locations of samples within the plate.


Real time detection of registration errors saves CM time and resources

LIMS Comparison feature also used to detect formatting errors

Using Sample color to track CM processes & sample quality

We now can query and identify color inconsistencies across entire compound collections or the same sample distributed across multiple working copies of

the library.

Plate Auditor data can be used to monitor consistency of sample color over time…



Why should we care about sample color?


Two copies of plate prepared by external vendor reveal inconsistencies


Running color comparison reveals color inconsistencies…

Sample 1

Sample 2

Sample 3

Sample 4

Sample 5


Color data identifies problematic samples earlier in CM processes

Sample color is indicative of compound quality (MLSMR)


Degraded sample (left) vs. expected sample (right), both verified via LC-MS

Expected sample (left) copy 2 vs. failed sample (right, precipitated) from copy 1, both verified via LC-MS

Expected cherrypicked samples (left, middle) vs. failed screening deck sample (right), all verified via LC-MS

Plate Auditor results predict LC-MS results

2011 vs. 2009

2011 vs. 2011

Cherrypicked samples vs. screening deck

Sample color is predictive of its performance in HTS


Typical primary screen plate

Typical confirmation screen plate

Why else should we care about sample color?

With the Plate Auditor, can we show a true correlation between sample color and its hit activity?

Sample color is predictive of its performance in HTS


Is there correlation between sample color and various assay properties?

ABSBLA

CALCIU

MCNGC

EPIA

BS

FLIN

T

FLUO8 FP

HTRFLU

MI

QFRET

TRFR

ET0%

10%

20%

30%

40%

50%

60%

70%

80%

% of hits containing color vs. HTS readout technology

N=90 assays

Readout Technology Average % hitsOverall average % hits

Sample color data has impacted HTS assay development (biologists) & follow up of HTS hits by Project Chemists

MLPCN HTS compound collection

HTS hits0%

2%

4%

6%

8%

10%

12%

14%

16%

18%

% Colored samples 12/2007-12/2011

Precipitate results measure effectiveness of solvation SOPs


Tubes containing DMSO solvated samples (pre-sonication)

Tubes containing DMSO solvated samples (post-sonication)

How can we monitor changes in precipitate? What impact does it have on CM operations?

Quantitative precipitate monitoring improved compound solvation SOPs

Heat, mixing solvent change, sonication, etc…

Using precipitate detection to monitor library health

Precipitate filled screening plate“Normal” screening plate

Plate Auditor automatically identifies plates with high levels of precipitate




Plate Auditor helps to identify solubility issues when analyzing samples from sublibraries …

Library% of library containing precipitate

Sublibrary A 0.1%

Sublibrary B 2.5%Sublibrary C 0.8%

A

B

C

… providing a useful dataset when communicating

problems with suppliers, collaborators and CM staff

internally.

BioFocus comparison: HTS copy 3 (in use) vs. HTS copy 4 (from freezer)

Plate # Copy 3 Copy 42 10 343 17 604 34 975 65 1286 24 39

TOTAL 150 358% increase 139%



Helps identify existing samples that need re-solvation...or remediation

Baseline precipitate levels can be established with Plate Auditor data

Representative plate from copy 3 of sublibrary B

Representative plate from copy 4 of sublibrary B

Interesting Observations: Labware QC


Why do we care if there are air bubbles in the polypropylene?Air bubbles cause plates to be warped & unusable with automation.

When creating new compound plates, Scripps discards plates with air bubbles as precautionary measure.



Air bubbles can also exist in 1536 well polypropylene plates.

In addition to plate warping, bubble artifact in 1536 well poses additional problems for machine vision QA/QC.



Plate Auditor images can be used to document plate damage and deformities


Are all wells created equally?


2.25mm

2.25mm

.28mm offset

.28mm offset

Plate with offset wells Plate with no offset wells

Plate from different vendor

What impact do offset wells have on experiments?



Plate Auditor provides data to communicate problems with third parties

Impact Review


Review of Plate Auditor impact at Scripps Florida

2) Using Sample color to track CM processes and sample quality:• Color data identifies problematic samples earlier in CM

processes• Plate Auditor results predict LC-MS results3) Use sample color as a predictor of sample performance in HTS:• Susceptibility of a HTS assay to color interference is

considered during assay development (biologists) and hit-follow up (chemists)

• Correlation between color and hit performance helps determine HTS false negative/positive rate4) Use precipitate as a measure of solvation effectiveness:

• Quantitative precipitate monitoring improved compound solvation SOPs

1) Detect LIMS database errors in real time:• LIMS DB errors are detected & corrected immediately upon

plate creation

5) Use precipitate as an indicator of library health:• Helps identify existing samples that need re-solvation...or

remediation6) Other interesting observations:• Images can be used to document plate damage and

deformations• Images can be used to communicate problems with third

parties



• Degradation of samples over time




Impact of Plate Auditor at Scripps

• Received samples are quickly QC’d upon arrival; Plate Auditor results are then compared to the vendor’s plate map.

• Plate Auditor periodically audits compound library over its lifecycle

• Automated QA/QC of samples removes subjective analysis and prevents CM staff burnout

• Plate Auditor plate “snapshots” allow us to easily & visually track sample genealogy

• Routine QC of large libraries now possible by automating labor intensive visual inspections


Plate Auditor Summary & More Information

Scripps has licensed the Plate Auditor technology to Brooks Life Science Systems for commercialization & sale. The Brooks Plate Auditor will be part of the Brooks sample analysis line of instrumentation.

Additional information, including Plate Auditor Pilot Screen results, can be found in “Monitoring of HTS compound library quality via a high-resolution image acquisition and processing instrument”. Baillargeon P, Scampavia L, Einsteder R, Hodder P. ; J. Lab Automation, June 2011


Lead Identification:Peter HodderPierre BaillargeonPeter ChaseJoseph DatskoChristina EberhartImarhia EnogieruRoss EinstederKatharine EmeryVirneliz Fernandez-VegaLina DeLucaFranck MadouxBecky MercerLaura Pedro-RosaChristine PinelloLouis ScampaviaTimothy Spicer

http://hts.florida.scripps.edu [email protected]

Acknowledgements

Scripps Florida Funding

Corporation


http://www.mli.nih.gov/index.php