ERROR in parallel process - after finish commands of genetrees and -fgt

[david6git]

Hello OrthoFinder team,

Thank you for the development of your tool!!
I enjoy it and I am trying to optimize my resources for several analyses.

After several trial/runs, I edit config.json:
iqtree3 -s INPUT -mset Q.MAMMAL -m MFP --prefix PATH/IDENTIFIER

Now, I am using a new set of species:
orthofinder -y -M msa -S diamond_ultra_sens -A mafft -T iqtree3 -t 36 -f $core --save-blast-commands

It stopped (MEM limit) and I resume from BLAST (it was finished):
orthofinder -y -M msa -S diamond_ultra_sens -A mafft -T iqtree3 $results -t 64 -fg $resultsfromcore

I run out of time in the cluster and see that it stopped in gene trees step:
So trying to resume:

1. I looked at genetrees OG.txt and iqtree.log files, OTF3 assigned 1 thread/gene to iqtree. 3 heavy OGs OG000000, OG000001, OG000003 were unfinished...
   So, I run 3commands manually (-t36 =10min) using the command in log --> rename OG*.treefile to txt and move trees to WorkDirectory/Trees_ids
2. I also run manually the SpeciesTree command too.
3. Edit the log.txt on $resultsfromcore --> showing WorkingDirectory_Trees: $resultsfromcore/WorkingDirectory
   Issues:
4. not found SpeciesTree --> I rename/move SpeciesTree.treefile to SpeciesTree_unrooted_ids.txt and make a rooted version too.

So I am trying -fgt as:
orthofinder -y -M msa -S diamond_ultra_sens -A mafft -T iqtree3 -fgt $results -s $mi_especies_tree -t $threads
Initially failed but I used -s with new Speciestree and now it seems to proceed.

Now, it is running but show this ERROR:
Reconciling gene trees and species tree

2026-04-10 17:19:55 : Starting reconciliation and orthologues
2026-04-10 17:19:55 : Inferring orthologues from gene trees
ERROR in parallel process, exiting.
NoneType: None
None
ERROR: An error occurred, please review the error messages they may contain useful information about the problem.
in 2.298820s
Processing... ━━━━━━━ ⠧ 6/16172 0:48:34
Could you help me to understand this ERROR and understand what happened. Did I miss something?

A second question: If I set up -t64 -a8. are a the number of cores for BLAST, MSA and tree tools?

Thank you very much for your support.
Best regards

[Jonathan-Holmes-Bioinformatics]

Hi @david6git

A couple of quick questions which version of orthofinder are you using? My guess is v3<. Secondly how many species are you trying to run and are you using a HPC?

We have removed many of the checkpoint options from version 3.1.3 onwards because of the complexity of allowing --core --assign to work with the additional checkpoints. I think what might be happening is an issue with the restarting from trees either due to the -fgt or the manual creation of the tree files, I'll have to investigate this. If you're having issues with scalability (memory/time problems) I'd recommend using --core --assign, you already have a species tree so you should be able to select a good number of representative species.

Also IQ-Tree is very memory hungry and as is infer orthologs so be careful if you are running a large volume of proteomes. The command -t will determine the number of BLAST, MSA and Tree Tools while -a determines the cores for orthofinder specific processes.

Let me know if you need anything else answered I'll relook at the -fgt checkpoints but as mentioned we've depreciated alot of them and most only work during core runs.

[david6git]

Dear Jonathan,

Thank you for your answer.
I am using orthofinder v3.1.2 in a HPC. (total 29 individuals)

Part A. core (core -f looks OK with checkpoints)
I would like to obtain single-copy-OG from core set. Later, use single-copy-OG to estimate divergence times.
I set up diamond_ultra_sens, mafft and iqtree trying to increase accuracy as I work at subspecies and breed level.

However, if I setup all available threads (-t 32 or 64 depending on node) in orthofinder, I have MEM error in iqtree gene_tree or it takes 2 weeks when -t 16 and 1.2Tb RAM.
Question: If using

Part B. Using all samples with -f or using --assign?
I could not resume from --assign, so I tried all samples in -f (my total is small).
On my 1st run (1st post), I included all samples in -f (n=29), so I can resume from checkpoint it I have ERROR MEM:
OrthoFinder version 3.1.2.

Reconciling gene trees and species tree
2026-04-10 18:21:07 : Starting reconciliation and orthologues
2026-04-10 18:21:07 : Inferring orthologues from gene trees
ERROR in parallel process, exiting.
NoneType: None
None
ERROR: An error occurred, please review the error messages they may contain useful information about the problem.
Processing... ━━━━━━━━━━━━━━━━━ ⠙ 4/16172 0:00:00
Processing... ━━━━━━━━━━━━━━━━━ ⠸ 4/16172 2 days, 15:53:21
So, I cancelled it as it looks stuck.

As a pull request (if possible):
It would be great if available to obtain the commands to make gene_tree and species_trees similar to -op. So, it can be run separately (like BLAST) and later supply them to orthofinder for the reconciling step. I tried to do it myself and use -fgt as I explained before but I guess did not success.

Thank you for your support.
David

[NikoLichi]

Hi @david6git and @Jonathan-Holmes-Bioinformatics,

I also second David's suggestion to have the option to run other tools like MAFFT and/or IQtree3 separately and then incorporate them in the orthofinder pipeline.

@david6git , did you manage to run the alignment and species tree successfully separately and then start OrthoFinder again? if so how?

I am using OrthoFinder:v3.1.3

Thanks in advance for any piece of advice,
Nicolas