Community DNA fingerprinting

Getting more out of Sanger sequencing.
The idea came from looking at Sanger sequence traces from heterozygote organisms. In these traces it was possible to identify the loci of heterozygocity and determine the two types present visually. We though and still wonder wether it would be possible to extract the species or type composition and possibly frequency from sanger sequencing of community amplicons.
This idea is currently shelved.
https://github.com/low-decarie/community-DNA-fingerprinting

Transcript

1. Getting more out of
   Sanger sequencing
   Etienne Low-Decarie
   Corey Chivers
   Community
   DNA
   fingerprinting
   

   View Slide

2. Microbial community composition
   (phyto)plankton
   images: NHGRI and NASA
   soil
   human microbiome experimental ecology and evolution
   

   View Slide

3. Community fingerprint
   ›  Determine
   community
   composition from a
   community level
   characteristic that
   results from the
   proportional sum of
   characteristics of its
   constituent types.
   ›  Syn. community profiling
   3x ½ x
   2x
   

   View Slide

4. Pigment fingerprinting
   bbe FluoroProbe Manual Page 10
   DETERMINATION OF DIFFERENT ALGAE
   The division of chlorophyceae (green algae) shows a broad maximum of fluorescence at the 470nm
   LED, which is caused by chlorophyll-a and -b. The cyanophyceae (blue-green algae) have their
   maximum at 610nm due to the photosynthetic antenna pigment phycocyanin. Cyanophyceae also
   contain chlorophyll-a if there is low intensity at 470nm. This is due to the masking effect of the
   phycocyanin. Furthermore, the high peak at the 525nm region for the bacillariophyceae originates from
   xanthophyll fucoxanthin and for the dinophyceae from peridin. The maxima at 470nm are caused by
   chlorophyll-a and -c. In our last analysed group, cryptophyceae, a significant maximum can be found at
   570nm, which originates from phycoerythrin.
   It is obvious from the figure below that it is not possible to distinguish bacillariophyceae and
   dinophyceae by their “fluorescent fingerprints”. But it can be clearly seen that it is possible to
   distinguish five groups of algae: chlorophyceae, cyanophyceae, dinophyceae and bacillariophyceae,
   cryptophyceae.
   Additionally, it should be mentioned that sometimes the phycocyanin content per cell in cyanophyceae
   varies. Nevertheless, the average fingerprint can be used to differentiate this division.
   0
   0,2
   0,4
   0,6
   0,8
   1
   1,2
   440 460 480 500 520 540 560 580 600
   Excitation-Wavelength/1nm
   Relative Fluorescence Intensity
   scenedesmus
   chlamydomonas
   monoraphidium
   rhodomin
   chlorella
   scenedesmus fal.
   micractinium
   scenedesmus quad.
   scenedesmus subs
   scenedesmus ob.
   microcystis viridis
   synnechocochus
   nrc-1
   aphanizomenon
   cyclotella
   nitzschia
   synedra
   cryptomonas pl
   cryptomonas a
   cryptomonas
   ceratien
   peridinium
   Micro2
   Micro7
   Micro8
   Micro9
   Micro12
   Micro19
   Micro13
   Micro18
   Micro15
   Micro16
   Micro23
   Micro25
   Micro26
   Micro20
   Micro27
   Micro21
   Micro11
   Micro17
   Micro22
   Micro10
   Micro24
   Micro14
   Micro4
   Chlorophyceae
   Dinophyceae
   Bacillariophyceae
   Cryptophyceae
   Cyanophyceae
   The fluorescence intensities of the 5 divisions divided by the intensity of the LED and normalised to the maximum
   intensity of each division. In this measurement, several species of the mycrocystis cyanobacteria (abbreviation
   micro) were tested as well.
   images: Academy of Natural Sciences, BBE moldaenke and others
   

   View Slide

5. Physiological fingerprint
   ›  Sole carbon source utilization
   ›  BIOLOG eco-plates
   ›  Toxicological
   ›  Antibiotic
   ›  Toxin and metals
   

   View Slide

6. Chemical fingerprinting
   ›  Fatty acid
   ›  Isotope
   ›  Trophic structure signal
   images:Rebecca E. Drenovsky, Roger A. Duncan, Kate M. Scow,SPC ocean fisheries programme,
   

   View Slide

7. DNA fingerprints
   ›  Non sequencing based
   ›  Terminal Restriction Fragment Length
   Polymorphism (T-RFLP)
   ›  Denaturing Gradient Gel
   Electrophoresis (DGGE)
   ›  Ribosomal Intergenic Spacer Analysis
   (RISA)
   ›  Sequencing based
   ›  High-throughput sequencing
   images: Xia Zhou, Gregory W. Schmidt, Roberto Danovaro and Antonio Pusceddu,
   Stephan Gantnera, Anders F. Anderssona, Laura Alonso-Sáeza, Stefan Bertilsson
   

   View Slide

8. Chain-termination (Sanger) sequencing
   images: Helmut Kae
   NAGTCGTCCA
   

   View Slide

9. Heterozygote trace
   0.00
   0.05
   0.10
   0.15
   5 10 15 20 25
   datum
   value
   Homozygote A
   0.00
   0.05
   0.10
   0.15
   5 10 15 20 25
   datum
   value
   Homozygote B
   0.00
   0.05
   0.10
   0.15
   5 10 15 20 25
   datum
   value
   Homozygote A+B
   

   View Slide

10. Our naïve logical leap
    If we think it has not been done,
    chances are:
    •  Wrong search terms
    •  Can’t be done
    •  Shouldn’t be done
    

    View Slide

11. Two type community trace
    0.00
    0.05
    0.10
    0.15
    5 10 15 20 25
    datum
    value
    Species A
    0.00
    0.05
    0.10
    0.15
    5 10 15 20 25
    datum
    value
    Species B
    0.00
    0.05
    0.10
    0.15
    5 10 15 20 25
    datum
    value
    Community A+B
    variable A T C G
    type: any taxonomic level with common primers and distinguishing loci
    genotype, species, genus…
    

    View Slide

12. Two type trace
    

    View Slide

13. Quantification
    ›  Sum of square optimization on base ratio
    (Community - (quantity • Type 1 + quantity • Type 2))2
    

    View Slide

14. Simulation
    ›  Random species sequences
    ›  Sum species values for
    community
    ›  Vary
    ›  Number of species
    ›  Number of variable loci
    ›  Proportion of species available
    (sequenced) for fitting
    ›  Noise
    ›  …
    

    View Slide

15. Simulation
    Number of species in the community
    Number of variable loci
    •  Robust if:
    •  number of
    variable loci >
    number of
    species
    •  Better at predicting
    species with larger
    frequencies
    •  Can withstand 10%
    noise with little loss.
    

    View Slide

16. Quantification of visually aligned traces
    0.519 0.453
    R2>0.95
    

    View Slide

17. Traces are shifted
    ›  Alignment
    ›  Segment shift through 0 padding
    ›  Discontinuous/rough surface
    ›  Grid search
    ›  Very costly (time~ rangeparameters)
    ›  potential alternatives
    ›  human eye
    ›  differential evolution
    ›  particle swarm
    ›  integer programing
    0 0 0 0 0 0 0 0
    0 0 0 0 0 0 0
    0 0 0
    0
    0 0 0 0 0 0 0
    

    View Slide

18. Traces contain drift
    ›  Only segments can be aligned
    ›  Trace segmentation
    ›  Provides population with
    distribution as potential measure
    of error
    

    View Slide

19. Distribution of quantification with
    segmentation and alignment
    Segmentation size
    

    View Slide

20. Next steps
    ›  Get your expert opinion/feedback
    ›  many potential issues but worth pursuing?
    ›  …
    ›  Show proportion calculated=true
    proportion
    ›  Dilution series
    ›  Volunteers for trace data?
    ›  …
    

    View Slide

21. Acknowledgements
    BELL LAB
    FUSSMANN LAB
    LEUNG LAB
    etienne.webhop.org etienne.low-decarie@mail.mcgill.ca corey.chivers@mail.mcgill.ca
    Contacts and code
    Adam Herman
    https://github.com/edielivon/community-DNA-fingerprinting
    Thomas Bureau
    Genome Quebec
    

    View Slide