TEstrainer

[minhasbushra]

I am trying to follow the protocol for TE identification from Baril & Hayward 2022 for TE identification. I am using a custom library so I want to use the BEE script (TEstrainer) from earl grey (after I have completed the repeat modeler and repeat masker). But there are differences between TEstrainer for earlgrey and TEstrainer. Would it be ok to use TEstrainer outside earlgrey to complete the BEE step? Thanks

[TobyBaril]

Hi,
Thanks for checking out Earl Grey! You should be able to use the normal TEstrainer script. The only difference with the one specifically for Earl Grey is some paths and defaults that improved integration with the new pipeline.
Best wishes,
Toby

[minhasbushra]

Thanks for your reply. I understand, it is just that in the BEE process of the whole earl grey pipeline (i am looking at the codes), I am unable to find trimAI and EMBOSS as a part of "iteration" repeated 5 times (in TEstrainer for earl grey) as mentioned in the methodology of paper https://doi.org/10.1186/s13100-022-00263-5 (I know this is not the protocol paper for earlgrey). From what I understand, trimAI and EMBOSS should be part of the TEstariner for earl grey with BEE and repeating the process 5 times. Sorry i am just trying to understand, let me know if i this is not correct.

Thanks, Bushra

[jamesdgalbraith]

The process which repeats 5 times completely replaces the BEE process described in the manuscript of EarlGrey. Instead of using trimAl the current version of the pipeline uses a custom developed script which performs a similar function. Likewise, the function formerly performed by EMBOSS is now performed by a Python script developed specifically for creating TE consensus sequences. I hope this makes this clearer for you.

[minhasbushra]

Thanks, I got it. Also, i have another concern about the TEstrainer. In lines 165, 172, and 184, I had to change find ./${DATA_DIR} to find ${DATA_DIR} to make it work, otherwise, there was some path error and it crashed.

[minhasbushra]

Also, the previous TEstrainer was working fine with the above change. I ran the newer version and got this error.

Trimming run 1
sh: /dev/tty: No such device or address
0% 0:44=0s ltr-1_family-274.fasta Traceback (most recent call last):
File "scripts/TEtrim.py", line 73, in
og_con = SeqRecord(seq= align[0].seq.replace("-", ""), id=align[0].id, name=align[0].id)
AttributeError: 'Seq' object has no attribute 'replace'
2% 1:43=0s ltr-1_family-274.fasta Traceback (most recent call last):
File "scripts/TEtrim.py", line 73, in
og_con = SeqRecord(seq= align[0].seq.replace("-", ""), id=align[0].id, name=align[0].id)
AttributeError: 'Seq' object has no attribute 'replace'
4% 2:42=0s ltr-1_family-16.fasta Traceback (most recent call last):
File "scripts/TEtrim.py", line 73, in
og_con = SeqRecord(seq= align[0].seq.replace("-", ""), id=align[0].id, name=align[0].id)
AttributeError: 'Seq' object has no attribute 'replace'
6% 3:41=0s ltr-1_family-16.fasta Traceback (most recent call last):
File "scripts/TEtrim.py", line 73, in
og_con = SeqRecord(seq= align[0].seq.replace("-", ""), id=align[0].id, name=align[0].id)

[jamesdgalbraith]

Hi again, thanks for highlighting that issue, it was due to version difference in Biopython. I've introduced a switch and it should all work for you now.

[TobyBaril]

Hi,

The new BEE process implemented by TEstrainer does not use Trimal or emboss. It codes a different consensus generation using the majority rule consensus and trims based on alignments. This is all coded within TEstrainer. Doing this, we no longer rely on other programs and have better control over exactly how this works. The whole BLAST, Extract, Extend, process is encoded in TEstrainer, but it may not be immediately clear as most of the subprocesses have now be coded specifically and other programs are not used anymore. If you would still like to use emboss and trimal, I would suggest looking at Earl grey release v1.3 in this GitHub, and you will find the code in the earl grey script.

[minhasbushra]

thanks a lot, I got it now.