Failed to reconstruct 1 families

[walkyshitake]

Hello,

I am running OrthoFinder resulting data of 7 species with CAFE5.
My command line is: ./bin/cafe5 -t ultrametric_tree.txt -i gen_fam.txt.

I have run it three times to check consistency. The estimated lambda value is similar across runs, but at the very end CAFE5 reports:
"Starting reconstruction processes for Base model
Failed to reconstruct 1 families"

I have read the README and searched previous issues, but I may have overlooked relevant details.
My questions are:

1. How should I interpret this message and how does it affect the overall results?
2. Is there a way to identify which family is failing?

Thank you very much for your help. I look forward to hearing from you.

[benfulton]

If you look at the base_asr output file, you will see the reconstructed values for each node in the tree, for each family. One of those families should be missing or have -1's for all the values - that was the family that could not be reconstructed. It shouldn't have much of an effect on the overall results (assuming you have more than a couple of families), especially if your primary interest is the significant likelihoods, which are computed independently of the reconstructed values. It's unusual for CAFE to choose a lambda value that is incompatible with a family, but not unheard of. I'd be interested to know if there was something unusual about the family that failed - a particularly wide range of values, perhaps?

[walkyshitake]

Hello,

We tried the comm and grep commands to search for the missing family and -1 values. However, the results suggested that more than one family is missing when comparing the input gene family list with the base_asr.tre output. No -1 values were found in base_asr.tre.

Could you clarify whether "family" here refers to an entire gene family, or to individual family IDs from the input file?

In addition, would the estimated lambda value affect the family reconstruction? Ours is around 2, which seems relatively high compared to values I have seen in previous issues.