JacusaHelper for RRD

[kevinbi]

What functionalities are available in JacusaHelper for looking at RRD data? The documentation says that AddBaseChangeInfo, AddEditingFrequencyInfo, etc. are only applicable for RDD data. Why is this the case? Is there something I can run on RRD results to get a similar well-summarized output of number and location of putative A to G editing sites, for example?

Thanks.

[piechottam]

What functionalities are available in JacusaHelper for looking at RRD data? The documentation says that AddBaseChangeInfo, AddEditingFrequencyInfo, etc. are only applicable for RDD data. Why is this the case? Is there something I can run on RRD results to get a similar well-summarized output of number and location of putative A to G editing sites, for example?

RRD data are more complex than RDD data. That is why there is limited functionality in JacusaHelper.
Here, we focus on defining RNA editing sites by means of RNA DNA differences.
Calculating base change info and editing frequency is supported by AddBaseChangeInfo, AddEditingFrequencyInfo, respectively.
In RRD data reference (genomic DNA sequencing data (gDNA)) information is missing and therefore distinguishing allele specific expression from RNA editing sites becomes difficult, e.g.: given a site where
in both conditions the following base calls are observed:

• condition 1: observed base calls: A, G
• condition 2: observed base calls: A, G
  Without additional data, it is impossible to discriminate if this site
• is polymorphic with different allele frequency expression, or
  *an RNA editing sites with different editing frequencies.

One possible way to circumvent this would be to filter the set of RRDs against a database of know polymorphic positions.

The favorable option is to use matching gDNA sequencing data.
Given, gDNA, and RNA-Seq data for 2 conditions (e.g.: RNA1, RNA2):

1.) identify putative RDDs (JACUSA.jar pileup -a H:1 gDNA.bam cDNA.bam:
gDNA vs. RNA1 → pRDDs1
gDNA vs. RNA2 → pRDDs2
2.) identify RRDs:
RNA1 vs. RNA2 → RRD12s
3.) Intersect pRDDs1 and pRDDs2
gives you the total set of sites that are covered in gDNA and RNA-Seq
these sites are

Next steps depend on what you are interested in from RRD data.
You can intersect the corresponding sets with tools such as bedtools and continue working with RDD-like data – which are supported by JacusaHelper.

IMPORTANT!!!
When you are looking for sites that exhibit significantly different editing frequency between two conditions it is crucial to identify the set of putative RDDs with:

• JACUSA pileup and NOT!
• JACUSA call-*.
  Example:
  For some site 10 As are observed in gDNA.
  Condition RNA1 shows the following base composition: 5 As and 5 Gs – site will be in the set of putative RDDs for JACUSA pileup and JACUSA call-*.
  Condition RNA2 shows the following base composition: 10 As – this site will only be recognized by JACUSA pileup! JACUSA call-* ignores this site because it does not a variant site when comparing gDNA (10As) vs. RNA2 (10As)!

Hope that helps – but it is hard to explain.

[kevinbi]

Yes that helps. Thank you!

[rtasakis]

Hi guys,
Reading the answer to @kevinbi's question:
So, @mitschka when comparing RNAvsRNA, you would recommend that we don't call jacusa call-2, but jacusa pileup? Or have I misunderstood? In addition to what the manual says, what other options you would recommend to avoid/include for optimal RNAvsRNA analysis?

Thanks a lot,
Rafi