Alerax failed Inferring reconciliations above 7000 gene families

[Thomas Sauters]

Hello,

When running alerax, I am experiencing an error at the `inferring reconciliations` step when I use over a certain number of gene families. Every family can be run individually with alerax in this dataset, and I have reproduced the error with different groups of orthogroups. The failure happens somewhere between 7,000 and 10,000 orthogroups. Does anyone have an idea for what might be causing this error? Relatedly, is it okay to use/ concatenate the output from runs of alerax on individual genes instead of running it on an entire dataset?

My dataset has 36 species with trees constructed in iqtree.

The error in question:

`[00:36:24] [Reconciliation] Inferring reconciliations
alerax: /data/rokaslab/thomass/sec-fumigati/AleRax/ext/GeneRaxCore/src/trees/PLLUnrootedTree.cpp:152: static std::string PLLUnrootedTree::buildConsensusTree(const std::vector<std::shared_ptr<PLLUnrootedTree> >&, double): Assertio
n `treePointers.size()' failed.
[cn1617:1028982] *** Process received signal ***
[cn1617:1028982] Signal: Aborted (6)
[cn1617:1028982] Signal code:  (-6)
[cn1617:1028982] [ 0] /cvmfs/soft.computecanada.ca/gentoo/2023/x86-64-v3/usr/lib64/libc.so.6(+0x38790)[0x148979c63790]
[cn1617:1028982] [ 1] /cvmfs/soft.computecanada.ca/gentoo/2023/x86-64-v3/usr/lib64/libc.so.6(+0x867ac)[0x148979cb17ac]
[cn1617:1028982] [ 2] /cvmfs/soft.computecanada.ca/gentoo/2023/x86-64-v3/usr/lib64/libc.so.6(gsignal+0x12)[0x148979c636f2]
[cn1617:1028982] [ 3] /cvmfs/soft.computecanada.ca/gentoo/2023/x86-64-v3/usr/lib64/libc.so.6(abort+0xd3)[0x148979c4d4b2]
[cn1617:1028982] [ 4] /cvmfs/soft.computecanada.ca/gentoo/2023/x86-64-v3/usr/lib64/libc.so.6(+0x223d5)[0x148979c4d3d5]
[cn1617:1028982] [ 5] /cvmfs/soft.computecanada.ca/gentoo/2023/x86-64-v3/usr/lib64/libc.so.6(+0x31662)[0x148979c5c662]
[cn1617:1028982] [ 6] /data/rokaslab/thomass/sec-fumigati/AleRax/build/bin/alerax[0x512785]
[cn1617:1028982] [ 7] /data/rokaslab/thomass/sec-fumigati/AleRax/build/bin/alerax[0x5249fd]
[cn1617:1028982] [ 8] /data/rokaslab/thomass/sec-fumigati/AleRax/build/bin/alerax[0x47326e]
[cn1617:1028982] [ 9] /data/rokaslab/thomass/sec-fumigati/AleRax/build/bin/alerax[0x4774bf]
[cn1617:1028982] [10] /data/rokaslab/thomass/sec-fumigati/AleRax/build/bin/alerax[0x42b6f8]
[cn1617:1028982] [11] /data/rokaslab/thomass/sec-fumigati/AleRax/build/bin/alerax[0x42be66]
[cn1617:1028982] [12] /cvmfs/soft.computecanada.ca/gentoo/2023/x86-64-v3/usr/lib64/libc.so.6(+0x2394a)[0x148979c4e94a]
[cn1617:1028982] [13] /cvmfs/soft.computecanada.ca/gentoo/2023/x86-64-v3/usr/lib64/libc.so.6(__libc_start_main+0x85)[0x148979c4ea05]
[cn1617:1028982] [14] /data/rokaslab/thomass/sec-fumigati/AleRax/build/bin/alerax[0x4225c1]
[cn1617:1028982] *** End of error message ***
/var/spool/slurmd.cn1617/job8678202/slurm_script: line 87: 1028982 Aborted                 (core dumped) ${basedir}/AleRax/build/bin/alerax -f ${orgfile} -s ${tree} -p ${outdir} --gene-tree-samples ${specnum}`

[Stefan Flaumberg]

NB Seems to be related to this unresolved issue.

Hi Thomas,

Please, make sure that you are using the latest AleRax version, v1.4.0. In this case, given you are working with quite many families, you could try the --memory-savings option and see if it solves the problem.

> is it okay to use/ concatenate the output from runs of alerax on individual genes instead of running it on an entire dataset?

If you are not estimating any global parameters, such as species tree topology/rooting, relative speciation order (species tree branch lengths), highways of horizontal gene transfers, global/per-species DTLO rates, speciation rate gamma categories, and are only trying to reconcile input gene tree distributions with a fixed species tree using per-family DTLO rates, then each family is processed effectively separately.

So then yes, with  --species-tree-search SKIP and --model-parametrization PER-FAMILY, this a valid workaround.

Best,

Stefan