Loading presentation...

Present Remotely

Send the link below via email or IM


Present to your audience

Start remote presentation

  • Invited audience members will follow you as you navigate and present
  • People invited to a presentation do not need a Prezi account
  • This link expires 10 minutes after you close the presentation
  • A maximum of 30 users can follow your presentation
  • Learn more about this feature in our knowledge base article

Do you really want to delete this prezi?

Neither you, nor the coeditors you shared it with will be able to recover it again.


Make your likes visible on Facebook?

Connect your Facebook account to Prezi and let your likes appear on your timeline.
You can change this under Settings & Account at any time.

No, thanks

MLPA Data Analysis Theorie

Coffalyser.Net methodology and work flow optmisation

Coffalyser MLPA

on 21 October 2013

Comments (0)

Please log in to add your comment.

Report abuse

Transcript of MLPA Data Analysis Theorie

MLPA Data Analysis
Achieving high calling accuracy
Optimise capillary separation
Correct signal sloping
Sample specific variation
In order to achieve high calling accuracy, sample to sample variation needs to be brought to a minimum and a robust and standardized MLPA data analysis strategy is required

Recognition of aberrations should include evaluation of the intrinsic significance of the found results rather than only comparing probe ratio on an arbitrary level
Estimate probe reproducibility
The reproducibility of a probe can be translated into a 95% confidence range that can aid with result interpretation
Minimise experimental variation
Minimising differences between the reference and test samples results in a more accurate normalisation
Robust normalisation
Using each reference probe for a separate normalisation between sample and reference increases robustness
Probe signal intensities should be significantly large and of similar magnitude among all profiles to decrease variation
Predictable variation, such as a drop in signal to the length of the products can be corrected using computational techniques
Variation specific to certain samples may influence only part of the probes and can result in false calls
Abundance of probes as compared to the sample DNA guarantees all target sequences in the sample to be covered
Product creation is dependent on the presence of 50-80 nt target sequences
Amplification is necessary to make signals detectable using capillary electrophoresis
The MLPA reaction
The MLPA probe signals
Single primer pair ensures all probe signals are in the same range
Probe amplification bias requires comparison to reference profile
Probe product measurements are proportional to the amount of the target sequences present
Test sample
Normal diploid female
NF1 complete deletion
Visual comparison
Reliability by applying a single strategy
High throughput
Data organisation
Analysis of complex profiles
Automated analysis?
Data analysis steps
Quality control
Data interpretation
Data normalisation
Data sorting & grouping
Fragment analysis
Capillary electrophoresis
(raw data files)
Color format:
Ratio < 0.7
Ratio > 1.3
NF1 region
Expected sample ratio results
Normal range
NF1 region
Expected sample ratio results
Peak detection and quantification
Size calling
Allelic designation or peak recognition
Raw data profile
Genomic profile
Prevent normalising samples that clearly will not provide reliable results
Quality of reference samples is crucial
Prevent normalising of samples with too low DNA concentrations
Prevent normalising samples that are incompletely denatured
D-fragments should have
an equal signal as compared to the lig fragment
Q-fragment signals should be less than 1/3 of the ligation fragment signal
Brings multiple sets of data on a common scale by dividing through a common variable (‘reference probes’)
The results of data normalization are probe ratios (DQ)
Probe ratios of below 0.7 or above 1.3 are regarded as indicative of a heterozygous deletion or duplication, respectively
Chromosomal aberrations often span larger regions
Aberrations can be recognized more easily this way and probes targeting the same region may confirm each other’s result
Reference probes are expected to be relatively equal to each other (reference probe region)
Probe with the same genomic variation are expected to have ~equal ratio’s (NF1 HZ del)
Sample to sample variation
Reference probe region
Test probes region
Variation between the reference and test samples influence the relative signal intensities of all probes
Results in waviness / roughness of the profile making result interpretation very difficult
= test sample
= reference sample
= signal test probe
= signal reference probe
Division scheme
Use the same tissue type for all samples
For tumour cells try resection margins
Make working solutions for all samples of 50ng/ul
Sample selection
Use a single extraction and fixation method for all samples
Beware of extraction methods that may leave PCR inhibitors such as phenol/trizol remnants, SDS and Fe containing beads
High EDTA concentration’s may also inhibit the PCR
Extraction / fixation method
Perform initial experiment on a selection of possible reference samples
Estimate quality extraction / fixation proces
Create pooled reference sample from best of selection
Optimise capillary separation process with this data
Initial experiment
Strategically disperse reference samples
Positive reference can be used to confirm results
Add noDNA control on last position
Add a tube with 8 ul water to check for evaporation
Experimental setup
Devices are designed for sequencing and not for quantification of fragments
Larger signals are more reliable and are less sensitive to variation
Detection occurs with CCD plates that may risk overloading or bleeding
Signals are optimally in a very small range, basically between 10-40% of the maximum detection range
Capillary separation
Equal start DNA concentration results in more equal profiles
Allow sample stacking by injecting low concentration buffer containing the sample in capillaries with a higher concentration buffer
Mix max 1 ul MLPA product in 10-24 ul Hi-Di formamide with minimal amount of size marker
Injection mixture
Significant influence on the signal strength
May require empirical optimization
Influenced by changing the injection time and injection voltage
Longer injection time == lower resolution
Injection voltage less predictable
If all fails, concentrate or desalt samples
Electrokinetic injection
MLPA probe products are recognised by the length of a peak resembling length of the probe
Product signal may be the peak heights or peak areas
Peak areas reflect the amount of product best, but are more sensitive to peak artefacts leading to false positives / negatives
Peak shape is mostly influence by polymer choice, run voltage and oven temperature
Separation and detection
Rerunning of injection mixtures best done directly after initial separation
Fluorescent dyes decay in formamide
Adjust injection time (voltage) and run voltage
Rerunning & optimisation
Drop in signal that is proportional with the length of the MLPA product fragments
Injection bias
Signal to size drop
Test sample
Reference vs sample profile
Regression analysis
Normal range
Ratio results after slope correction
Data sorted on probe length
Longer probes look falsely deleted
Shorter probes look falsely gained
Result without slope correction
Performing multiple strategically distributed reference samples (normal diploid control) repeatedly in each experiment can serve a double role:
Can be applied to normalise each sample
Can be compared to each other to measure the reproducibility of each probe
Multiple references sample
Relative signals are expected to be ~equal to each other
Reference sample 3
Reference sample 2
Reference sample 1
Comparing reference samples
Probe ratios in reference samples are expected to vary around ratio 1
Normalised reference sample data
Heatmapping data may help to identify the extremes
The most variable probe
Probes are expected to behave according to the normal distribution
By calculating the standard deviation over the reference samples confidence ranges can be estimated
Standard deviations should smaller than 0.1
Confidence ranges can be displayed as boxplots in charts to aid in result interpretation
Standard deviation & confidence range
Exon 42
Exon 23
Exon 15
Normalised reference samples (4)
Box plots & normal distribution
Type of box plots
Box = 50%
Whiskers = 99%
Box = 50%
Swiskers = 95%
Adapted simple
Box = 95%
Default in Coffalyser.Net
95% confidence range
arbitrary range
NF1 region
Box plots & result interpretation

Differences in genomic structure may interfere with the probe hybridisation of targets
Differences in the constituents of the samples may influence the amplification efficiency of certain probes
Causes sample specific variation
Sample specific variation may be recognised by the differences over the used reference probes

The total amount of variation of for a sample can then be estimated by:
Experimental variation + sample specific viariation
Estimating total sample variation
Normalisation scheme multiple reference samples
NF1 region
95% confidence range
test sample result
95% confidence range
Distribution comparison
95% confidence range test sample
95% range reference samples
Arbitrary Normal range
High confidence (<<*)
95% confidence range test sample
95% range reference samples
Arbitrary Normal range
Ambiguous area
Low confidence (<*)
Estimated 95% confidence range test sample
95% range reference samples
Arbitrary Normal range
Ambiguous area
Ambiguous result (?)
Use statistical techniques to understand the relationship between the probe length and the signal intensity
Correlating these two variables can be expressed by a regression line that may be used to correct the probe signals for this bias
After correcting the signals no bias in signal to length is expected.
= Reference sample
= No DNA control
Sample plate setup
Polymer choice and run voltage
Coffalyser.Net methodology and work flow optimisation
Signals of the probes compared to those obtained from a reference DNA sample known to have two copies of the chromosome, are expected to be 0.5 times the intensities if a copy is deleted
Full transcript