Paolo Kunderfranco | 15 Feb 14:33 2013


Dear Lucas Chavez

I followed MEDUSA protocol to filter out both not properly paired, low
quality mapping and non-unique sequences from my alignment files to use
MEDIPS fur further analysis of DMR.

For example one mC sample started with 100 milions reads. 80 % mapped, 70 %
of them properly mapped with high quility (mapQ>40).
The problem arises when I filter out for non-unique reads. Roughly 90 % are
discarded leading to a final number of 2-4 milions of reads.
All my mC samples behave in the same way.

Maybe the DNA starting material was not properly quantified (2-3 ng instead
of 5 ng were used for the generation of the libraries).
We didn't observe the same problem for the Input DNA ( correctly
quantified) and for 2 samples out of 4 for 5-hydroxy-mC.

The high number of non-unique reads could be due to a technical problem or
a biological problem? Have you ever experienced a similar problem?
How do you think I should proceed with the analysis? Is it absolutely
necessary to remove non-unique reads for MEDIPS analysis?

Is the first time I deal with this kind of analysis I would like to
undestand which is the best approach to follow.

I tried to run MEDIPS.saturationAnalysis with the following samples and the
correalation looks fine:

[1] 1890528

[1] 1.890528e+06 9.997250e-01

[1] 9.452640e+05 9.994605e-01

Is it possible that such a low number of reads is sufficient to generate a
saturated and reproducible methylation profile?

Thank you very much for your time,

	[[alternative HTML version deleted]]

Bioconductor mailing list
Search the archives: