See how Xoma solved the problem of integrating data management with scientific analysis in antibody discovery with an award-winning application.

1. XAbTracker® & SeqAgent®: Integrated LIMS and Sequence
Analysis Tools for Antibody Phage Display
February 20, 2017
Mark Evans
Best Practices in Personalized and Translational Medicine Short Course

2. 2
 Established in 1981
 Located in Berkeley, CA
 Small, publicly traded biotech company
 Experts in antibody discovery, optimization, cell line and process
development
 Currently supporting ongoing Phase 2 clinical trials
• XOMA 358: congenital hyperinsulinism & Post-bariatric surgery hyperinsulinism
• XOMA 213: Various hyperprolactinemias
About Xoma

3. 3
 Antibody Phage Display technologies are well established
after more than 10 years of use in the Pharmaceutical industry
as important drug discovery tools.
Scientific Background
Phage library Combine phage + antigen
Wash
EluteAmplify
Assay
Sequence
Heavy chain
Light chain

4. 4
 What has not kept up is adequate data analysis and data
management systems.
 Screening, DNA sequence analysis and candidate selection
can still be very time consuming.
 We found that the data analysis aspect was a major bottleneck
 Limits the number of drug development projects the pipeline
could handle
Problem

5. 5
 Developed two integrated software applications
 SeqAgent™ - integrated DNA sequence analysis package
specifically designed for use with antibody V-regions (Fv)
• Semi-automated pipeline
• Input is zipped DNA sequence files
• Converted to protein sequence
• Identifies framework and CDR structural features
• Produces protein sequence alignment
• Highly annotated and ready for final analysis.
Solution

6. 6
 Developed two integrated software applications
 SeqAgent™ - integrated DNA sequence analysis package
specifically designed for use with antibody V-regions (Fv)
 XAbTracker™ - a clone / assay data management system.
• Tracks clones throughout discovery process
• Tracks and evaluates associated assay results
• Integrated sequence identification via SeqAgent™
• Provides flexible workflow and data management for antibody discovery
Solution

7. 7
 Heterotetramers
 8 Constant and 4 variable regions
 16 light chain families
 7 heavy chain families
 Variable region
• 4 conserved framework regions
• 3 hyper-variable regions
Specific Challenges: Problem of data complexity
VL
CL
VH
CH1
CH2CH3
CH2CH3
CH1
VH
CL
VL
Fv
Fab
Light Chain
Heavy
Chain

8. 8
SeqAgent™ analysis workflow
Upload compressed
DNA sequences
Evaluate sequence quality
Identify low-quality regions
Translate proper reading frame
Compare protein seq against profile HMM
Identify light and heavy chain families
Identify constant and variable regions
Specific sequence pattern recognition
Cluster HC and LC sequences by
Levenshtein algorithm
Assemble and annotate display viewDisplay analysis result
UserActivity
ServerActivity

9. 9
SeqAgent™ Results Display
1
2
3 4 5 6 7 8 9 10
11
1) View management. Add / remove additional
sequences or copy view.
2) View controls.
3) Sequence selection box
4) Unique coded sequence identifiers.
5) Heavy and Light chain bin identifiers. Closely
related chain sequences have the same bin
identifier.
6) Unique light and heavy chain sequence
identifiers.
7) Representative box. Select sequence to
represent a bin.
8) Example of additional tags, signals, etc that
are automatically identified.
9) Framework and CDR regions are identified
and color coded. Alignment gaps are
indicated by a dash.
10) Poor DNA sequence quality glyph.
11) Grouped rows that have the same color
background indicate identical chain sequence

10. 10
 Low quality sequence as well as stop codons, potential post
translation modification sites are indicated on the sequences
SeqAgent™ Results Display
1
2
1) Showing Query-anchored view, the first row is the anchor for the bin and
the second row is identical.
1) Additional glyphs indicating post translational modification site and an
amber stop codon.

11. 11
SeqAgent™ Results Display
• Individual sequences can be inspected
• Sequences of light and heavy chains are tracked as paired sets which
represent functional antibodies

12. 12
SeqAgent™ Results Display
• Details about individual
chains may still be
accessed

13. 13
 Tracking large numbers of clones, replicates, assays, rearrays,
etc. is no trivial task
 100s to 1000s of individual bacterial colonies are picked into
96 well plates for screening.
 In addition to sequencing, clones are assayed via ELISA,
FACS or SPR methods.
 In most cases, the original raw data file is parsed directly into
XAbTracker™, with the exceptions coming in as tab-text after
preprocessing elsewhere, and is associated with the correct
clone.
XabTracker™ Data Management System

14. 14
 Which libraries are used
 What the target is
 Which antigens are being screened in each assay
 Organizational concepts such as Projects, Studies, Study
Rounds, Screens and Assays
 Several unique naming conventions
• Individual heavy and light chain sequences
• Antibodies and their format (IgG, Fab, scFv)
• Individual clones, reformatted clones, engineered clones
XabTracker™ keeps track of…

15. 15
1.Set thresholds for all plates
dynamically to data set min and
max values
2.Button locks the results of this
analysis in the database
3.Total hit indicator for each plate
4.Data quality (histogram/ scatter
plot).
5.Per plate thresholds can
override the master threshold.
6.Threshold line indicated in blue,
hits in red. Graph values
dynamically linked to the plate
view, hit colors automatically
reflected in the assay plate.
7.Wells in the plate view are
colored in shades of blue
across 11 scaled bins to
indicate data diversity range.
When they are a hit, they are
colored on a red scale to
indicate magnitude of the hit.
XabTracker™ 1
23
4
5
6
7

16. 16
 Multiple antigens and/or
multiple analysis criteria can be
part of an assay
 Two different antigens and a
total of three different analyses
are summarized.
 Red, orange and green in the
pie chart legend indicate which
pie slice will show the analysis
result for that assay.
 Hits are indicated in yellow and
blue is a non-hit
XabTracker™ Analysis
Summary View

17. 17
XabTracker™ Improvements
 Interactive decision making using
Venn diagrams

18. 18
 Integrated 3D
structure prediction
pipeline and display
XabTracker™ Improvements

19. 19
 MacPro server
 Django / Apache / PostgreSQL / Python /Jquery / D3 stack
 Dependencies
• Phred / Phrap
• Hmmer 3
• Emboss
• ClustalW
• BLAST
Technical details

20. 20
 Developed in-house
• Derived from open-source resources
• Less than a year by a team of three
 Number of samples that can be analyzed increased >10x
 Analysis time: days / weeks  minutes / hours
 Standardized analysis methods allow consistent data
interpretation
 Prosecution of drug targets per year has increased 3x
 A significant ROI on the manpower costs and minimal cost
• (< $12K) of commercial license fees needed for access to certain open
source libraries.
ROI

21. 21
 Laboratory software is often plagued by antiquated interfaces
 We have developed a relatively lightweight, nimble data management and sequence
analysis application suite that is specifically designed for antibody discovery
• As thin client systems, they are able to run in web browsers.
• Since they utilize responsive web UI components, the applications work equally well on
PC, tablet and even smart phone platforms, providing the users with maximum
flexibility.
 We believe that these applications provide a good illustration of what the future of
laboratory software will look like
Conclusions

22. 22
 For inquiries contact Zander Strange
• zander.strange@xoma.com
 Thanks to
• Yevheniy (Eugene) Chuba
• Matthew Batterton
• Lauren Schwimmer
• The Discovery Research group at Xoma
• BioIT World and CHI for providing this opportunity
Finally…