Problems with cactus-update-prepare

[mufernando]

In following the instructions described in https://github.com/ComparativeGenomicsToolkit/cactus/blob/master/doc/cactus-update-prepare.md#replacing-a-genome, I ran into some issues:

1. cactus-blast and cactus-align complained about the --realTimeLogging flag, which (for these two commands) needs a --realTimeLogging True (but note that cactus-preprocess doesn't need the boolean).
2. cactus-align complained about multi-furcations. So I copied the XML from

   cactus/src/cactus/cactus_progressive_config.xml

   Line 470 in fe99fe1

   <!-- allow_multifurcations: added in v2.7.2 and defaulted to 0, since multifurcations (ancestors with > 2 children) have been shown to reduce coverage. -->

   and set allow_multifurcations="1".

I would be happy to contribute a PR modifying the python script cactus-update-prepare to add the boolean after --realTimeLogging for the two commands that need it.

The multifurcations issue is a bit trickier to deal with. Do we require that cactus-update-prepare be passed a custom XML config (to be used by cactus-blast and cactus-align?

[glennhickey]

Thanks. For --realTimeLogging the best is just to not bother setting the flag at all (it's on by default).

For the multifurcations, that is indeed trickier. The current version of cactus is not as robust to them, and it's best to avoid them. What step exactly does it want to make a mulitifurcation for? It seems like it shouldn't be necessary when replacing a genome.

[mufernando]

Hi Glenn,

1. cactus-blast and cactus-align seem to expect the boolean value after --realTimeLogging, that is I get an error with the current implementation that doesn't have this.
2. I am adding a genome to a node (which already had two child nodes), so multifurcation is necessary. I want to have two different releases of one of the species genomes.

Thank you,

Murillo

[glennhickey]

1. If you make a PR to remove --realTimeLogging altogether that would be best.
2. You are best to add to a branch instead, since the current version of cactus does not properly support multifurcations -- they can lead to large coverage drops in some cases.