Refining families...Refiner: No consensus could be derived from the input sequences

[maggiesudo]

Hi team,

I have been using RepeatModeler on multiple different plant species and recently came across this issue. Do I need to be concerned? The RepeatModeler seemed to be able to process other families successfully. I only got error like this for one or two family most of the time.

LTR Structural Analysis

Running LtrHarvest... : 00:31:55 (hh:mm:ss) Elapsed Time
Running Ltr_retriever... : 00:22:46 (hh:mm:ss) Elapsed Time
Aligning instances... : 00:34:20 (hh:mm:ss) Elapsed Time
Clustering... : 00:04:43 (hh:mm:ss) Elapsed Time
Refining families...Refiner: No consensus could be derived from the input sequences in /home/proj_202504p2_BD-MED/maggie/project/SG_diversity/assemblies/SG_Tain/results/repeat/RM_1447042.WedApr151957462026/LTR_974953.ThuApr161929072026/ltr-1_family-139.fa
Refiner: No consensus could be derived from the input sequences in /home/proj_202504p2_BD-MED/maggie/project/SG_diversity/assemblies/SG_Tain/results/repeat/RM_1447042.WedApr151957462026/LTR_974953.ThuApr161929072026/ltr-1_family-141.fa
: 00:21:44 (hh:mm:ss) Elapsed Time
Program Time: 01:55:30 (hh:mm:ss) Elapsed Time
-- Clustering results with previous rounds...
- 1252 RepeatScout/RECON families
- 191 LTRPipeline families
- Removed 59 redundant LTR families.
- Final family count = 1384
LTRPipeline Time: 01:56:39 (hh:mm:ss) Elapsed Time

Thank you and I hope to hear from you soon.

[rmhubley]

This can happen for a number of reasons including unalignable sets of TE instances produced by the de novo algorithms. Obviously those algorithms shouldn't be producing unalignable sequences, but it's not always obvious what went wrong. I would be happy to do a deeper dive into this if you are able to provide the assembly, the version details of the software you used, and the random seed (printed at the start of the run) that was used.

[maggiesudo]

Unfortunately our assembly is not published so I am not allowed to share it. I can provide the fasta files of the affected families though. Would that be useful? They look like unalignable sets of TE to me.

The RepeatModeler version is 2.0.7 and the random seed used was 1776254266. The genome size 802MB.