A-I editing

[Arezzo99]

Script used Reditools3 for RNA:
Hi, please can you confirm that if studying A-I editing, I should only use the AG allsubs (Reference-A) sites and not use TC at all in further analysis? Also is there any filtering for no subs in reditools, as I am wondering whether any sites that show editing frequency of 0 are lost (i.e. upon enzyme kds).

How does reditools3 compare to reditools1 and 2 for doing this?

[ahanden]

Hi, please can you confirm that if studying A-I editing, I should only use the AG allsubs (Reference-A) sites and not use TC at all in further analysis?

If you are doing a paired end analysis and use the --strand option set to 1 or 2, you only need to look for AG in allsubs. Otherwise, you need to look for both AG and TC, as you won't be able to differentiate between strands.

Also is there any filtering for no subs in reditools, as I am wondering whether any sites that show editing frequency of 0 are lost (i.e. upon enzyme kds).

If you use --min-edits 0, the results will include positions with no editing events.

How does reditools3 compare to reditools1 and 2 for doing this?

All three versions use the same logical steps to detect and determine editing events. Version 2 requires less setup and input as compared to version 1. Version 3 is significantly faster than version 2 and requires even less setup. You should get the same results regardless of which version you use.

[Arezzo99]

Many thanks for your reply. Please can I confirm that if i have a forward stranded approach we should also be be running -C as well -s2 the parameters. Can you confirm why/if the -C is necessary?

Given a kd file may not have AG anymore or any edited bases surely there should be no subs and -me and -men as 0 applied otherwise you may miss sites with complete loss of editing

[ahanden]

Many thanks for your reply. Please can I confirm that if i have a forward stranded approach we should also be be running -C as well -s2 the parameters. Can you confirm why/if the -C is necessary?

In REDItools v3, -C, in combination with --strand 1 or --strand 2, will filter reads to make sure all results are on the same strand. For example, if you have 5 reads on the forward strand and 3 on the reverse, -C will only report the 5 reads on the forward.

Note: this is one of the few options that differ between REDItools versions. Versions 2 and 3 treat -C the same. Version 1 requires -C in order to report the base complement on the reverse strand (which versions 2 and 3 do automatically). There are plans to change REDItools3 to behave the same way as version 1 for this option, but it hasn't been implemented yet.

Given a kd file may not have AG anymore or any edited bases surely there should be no subs and -me and -men as 0 applied otherwise you may miss sites with complete loss of editing

You are correct. Conveniently, -men is zero by default, so you shouldn't need to set the option.

[Arezzo99]

1. Hi thanks a lot. Regarding the -C in REDItools3, does this mean we need to apply -C in combination with --s2, otherwise it will include unedited and edited base counts from the wrong strand too (not be reflective of editing)? Please can you explain what you mean by the differences between REDItools1, 2 and 3 outputs in regard to -C, as not sure I fully understand.

2. REDItools3 i thought already automatically converts based on strand all the T to A when running it on stranded sequencing (--s2) data. Therefore if running REDITools3 on forward stranded RNA-seq data (--s2) we should only take the sites that reditools3 gives AG outputs for differential A-I RNA editing analysis and ignore TC subs completely?

3. While -men default is 0, the -me default is 1, which means you won't pick up 100% editing lost sites right? So you need to actively set -me as 0 too.
   Is there any other parameter change you would do differently from default other than --s2,C, and -me as zero for Differential RNA editing analysis on forward stranded RNA-seq data?

4. Lastly, Is there a REDItools3 equivalent of annotate_with_DNA.py from REDItools2 for annotating RNA sites with DNA information, or is using the REDItools2 script still the recommended approach?

Thanks so much for your kind help, much appreciated

[ahanden]

1. Hi thanks a lot. Regarding the -C in REDItools3, does this mean we need to apply -C in combination with --s2, otherwise it will include unedited and edited base counts from the wrong strand too (not be reflective of editing)? Please can you explain what you mean by the differences between REDItools1, 2 and 3 outputs in regard to -C, as not sure I fully understand.

Yes, I recommend using -C in combinations with -s.

If you use -C and -s, you will get the same results for all versions of REDItools. If you use -s without -C, version 1 output will be different from versions 2 and 3.

2. REDItools3 i thought already automatically converts based on strand all the T to A when running it on stranded sequencing (--s2) data. Therefore if running REDITools3 on forward stranded RNA-seq data (--s2) we should only take the sites that reditools3 gives AG outputs for differential A-I RNA editing analysis and ignore TC subs completely?

Yes, that is correct. Using -s with REDItool3 (latest release v3.6 and back) will report T-to-C on the reverse strand as A-to-G, so you can ignore TC's.

3. While -men default is 0, the -me default is 1, which means you won't pick up 100% editing lost sites right? So you need to actively set -me as 0 too.
   Is there any other parameter change you would do differently from default other than --s2,C, and -me as zero for Differential RNA editing analysis on forward stranded RNA-seq data?

In order to report on all locations for which you have reads (i.e. sites without edits), you will need to use -me 0.

There aren't any other parameters you need, but there are a couple parameters you may find useful for your analysis. Many people set -l/--min-read-depth to 5 (or more) to increase the confidence of the result. Since you are only interested in A-to-I edits, you might want to use -e/--exclude-multis to remove any locations that have multiple variants

4. Lastly, Is there a REDItools3 equivalent of annotate_with_DNA.py from REDItools2 for annotating RNA sites with DNA information, or is using the REDItools2 script still the recommended approach?

REDItools3 has an annotate tool that does the same thing as annotate_with_DNA.py from REDItools2, and is marginally faster. python3 -m reditools annotate

[Arezzo99]

Hello,

I assume we should be running ---dna parameter when running reditools3 on WGS data too?

Secondly, is there a bug, since tried running -me 0 on reditools3 v3.6 and get this-

[ERROR] (<class 'ZeroDivisionError'>) division by zero
[ERROR] Killing job

[Arezzo99]

Since in a previous run v3.2, we didn't change the -me from default and we saw 0 edited base sites so wondering whether that was a bug at the time or whether the default was 0 then

[ahanden]

I assume we should be running ---dna parameter when running reditools3 on WGS data too?

Yes, use the --dna parameter.

Secondly, is there a bug, since tried running -me 0 on reditools3 v3.6 and get this-

[ERROR] (<class 'ZeroDivisionError'>) division by zero
[ERROR] Killing job

Yes, this is a known bug. It has been addressed in our development branch, but we haven't made a release with the fix. You can clone the REDItools repository (git clone https://github.com/BioinfoUNIBA/REDItools3.git) and run REDItools from inside the repository to get around the issue.

[Arezzo99]

Hi, I wanted to check whether we can do a form of de-novo analysis by seeing sites edited upon gene modifications/disease but not edited in control files just by using the output of reditools3?

[ahanden]

Yes, you can do a comparison of RNA editing between two RNAseq datasets.

You need to use the analyze tool on each NGS dataset, separately. Then you can use the annotate tool to compare your control run to your disease/gene modification run.