MedIP-Seq and Batman

[luqie]

Anyone knows how to use Batman for MedIP-Seq?

I've been emailing the author for months without reply and its very very disappointing. The accompanying document tells us to contact the author on how to do it, but this seemed like a fruitcake.

[Layla]

any luck?

Hi lugie

Are you still stuck with this?

L

[luqie]

Hi Layla,

sorry for the late reply! Thomas, thru Vardhman, the 2 authors of Batman, replied and helped me out with some troubleshooting.

I already have the full recipe to input the Medip-Seq data into Batman but i need to do some pre-processing first in Maq and some perl scripts.

Are you doing this as well?

[pko]

Please note, both the manual and the paper have an outdated e-mail address of this software author.

[node]

Hi here is GentleYang from BGI.
We also use batman in our "medip" but with something wrong. When we load probes data（GFF）into our database ,the batman.LoadProbesApplication alway throw a exception like multiple keys into table medip_roi !

Is every line with uniq ROI id ?

[lijuntan]

Hi, Luqie

I am going to use Batman for MedIP-Seq.I will soon get Medip-Seq data. I am a novice. I have no idea on how to use it. Can you help me or provide some reference to me ?

[pzs]

General guide?

I was planning to use Batman for MeDIP-seq too - are there any general guides anywhere? Or an alternative tool?

[sunsnow86]

Luqie: could you share some of your experience in using Batman? what is the recipe for Medip-Seq analysis. Thanks

[haidan]

Anyone knows how to use Batman for MedIP-chip?

When I used batman.CalibrateApplication. It always tells me that Baseline=NAN Response=NAN and the scatter.dat,trend.dat have no any contents. I'm really want to know the reason.

[alkalait]

When I used batman.CalibrateApplication. It always tells me that Baseline=NAN Response=NAN and the scatter.dat,trend.dat have no any contents. I'm really want to know the reason.

I have the exact same problem. Have you solved it?

[chitra24]

Hi,
I am analyzing medip-chip data. Please could any of you help me with some issues

I wish to identify differentially methylated regions (DMR's )as the final goal.

Which software/ pipeline is good for this? I have used bioconductor Ringo package for determining regions enriched in methylation. Ringo is designed for chip-chip data analysis, but is it still applicable for medip-chip?
It is right to identify DMR's from the peaks determined by ringo using some statistical test like t.test ?

Please kindly let me know you suggestions as soon as possible since im greatly confused with this!

[plombard]

Hi, I've got the same query again. I have some MeDIP-seq data and I am wondering if anybody has any guides or information how to run it?

[asurarocks]

Luqie: could you share some of your experience in using Batman? what is the recipe for Medip-Seq analysis. Thanks

Luqie: Please, help us!

[fczqx]

the recipe of batman

Hi Luqie, would you mind share your expriment of using Batman to us ? I recently using batman to anaylsis MeDIP data, but batman seems doesn't work.

Hi Layla,

sorry for the late reply! Thomas, thru Vardhman, the 2 authors of Batman, replied and helped me out with some troubleshooting.

I already have the full recipe to input the Medip-Seq data into Batman but i need to do some pre-processing first in Maq and some perl scripts.

Are you doing this as well?

[asurarocks]

I see a lot of people still looking for the protocol for using Batman. I contacted the author and he replied me with the protocol for using Batman on MeDIP-seq. I will just paste it here:
"Here's the current protocol. Give me a shout if you've got any questions.

One issue: there's an incompatibility between the current Batman code and recent versions of the mysql-connect/j database driver. One day I should fix this, but for now please use an old version. This one is known to work
well:


ftp://ftp.mirrorservice.org/sites/ft...a-5.0.5.tar.gz

Best wishes,

Thomas.


I've just put together a new Batman snapshot which includes a complete set of pre-processing scripts for handling MeDIP-seq data. You can get it here:

http://td-blade.gurdon.cam.ac.uk/software/batman/batman-20090617.tar.gz

This snapshot includes some minor known issues (the one that's likely to bite people is occasionally crashes when loading multiple MeDIP-seq datasets into a single database. For now, please create a new DB for every dataset), but I'm hoping to have a full release, and some updated documentation, ready in the next few days.


For now, the MeDIP-seq recipe goes something like:


1. Align your reads to the reference genome. I currently recommend MAQ for doing this. Other packages should work, but you'll need to write your own scripts to:


a) discard low-confidence alignments (if they're reported by your preferred aligner).


b) converting the tag placements to GFF2 format.


2. Create a Batman DB and register the experiment as normal.


3. Generate and load a suitable coupling-profile:


java -server -Xmx500M batman.EstimateCouplingProfileApplication -minLength 500 -maxLength 500
>ms500_coupling.txt

java -server batman.LoadCouplingProfileApplication
<db details> -name ms500 ms500_coupling.txt


4. Pre-process your aligned MeDIP seq data:


maq mapview medipseq.map | java -server batman.medipseq.MaqToGFFApplication >medipseq.gff


(You can concatenate several GFF files at this point if you're using multiple lanes of sequence for one sample).


for chr in `seq 1 22` X Y; do echo $chr; java -server -Xmx1800M batman.medipseq.ReadsToPseudoArrayApplication -target $chr -fragmentLength 200 -fragmentExtension 500 medipseq.gff
>medipseq-chr${chr}.pad.gff; done



The "pad.gff" files are in normal Batman array format (hence pseudo-array). This script automatically generates one ROI (called something like "CHR6" for each chromosome. You'll use these when generating methylation profiles).


5. Load the resulting "pad.gff" files into the Batman database as usual, using LoadProbes then LoadRatsGFF.


6. Calibrate using the appropriate coupling profile. Please specify one chromosome (doesn't really matter which one) when you do this:


java -server -Xmx1800M batman.CalibrateApplication <db options> -couplingProfile ms500 -expt medipseq -chr 6 -writeDB


7. Generate the methylation profile:


java -server -Xmx1800M batman.SampleMethStatesApplication -expt medipseq -couplingProfile ms500 -tilepathSamples CHR6 -mseqHack -proxySpacing 100 -tilepathTile 500 -tilepathStep 500 -tilepathFlank 500 -precision 0.25 2>batman-chr6.log | java -server batman.SumarizeApplication -trim >batman-chr6.gff


Obviously, you'll need to run this for every chromosome. If you have the computing power available, you'll probably want to analyze several chromosomes worth of data in parallel.


As always, give me a shout if you run into any problems.


Thomas"

Again, this was provided by Dr. Thomas Down, I just pasted it here for everyone's use. I hope it helps!