PF2 spinning at 'Using a separate GTR+G model for each data block'

[Teagan Mulford]

Hello all! 

I saw a similar issue without any resolution, so I figured I'd post my own.

I am running partition finder 2.1.1 on a cluster, with partitionfinder loaded as a module. I've tried running another dataset, and it worked well, but for some reason, with my data it keeps hanging on "Using separate GTR+G model for each data block ", and never produces any errors or other context. It gets killed after 2 days and produces no meaningful outputs.  

I'm using 10 UCE loci, chosen by SortaDate, but I can't find any meaningful differences between my data, and the dataset that works. I know I have some missing loci (they're from a 70% completeness matrix but I don't *think* that would be causing the error, especially as I have tried to run it with one gene with no N's and gotten the same issue). 

I tried rerunning with "module load partitionfinder / PartitionFinder.py 10gg -v -p 1" after reading the "Things that will help" section of the group. I'll attach that, and my config and phylip files to this post. 

Any help would be appreciated! I'm sure it's a dumb mistake on my end, but I really would like to figure this out! 

Thank you,

Teagan

[Rob Lanfear]

Hi Teagan,

Please switch to using IQ-TREE2 for your analysis. 

Details are here: 
http://www.iqtree.org/doc/Advanced-Tutorial

The command line you want will look something like this:
iqtree -s example.phy -p example.nex -m MFP+MERGE -rcluster 100

using -rcluster 100 will do the relaxed clustering algorithm but analysing 100% of the schemes, which is essentially the same as the greedy algorithm.

We have almost finished finalising all the details, but the main point is that everything that PartitionFinder does (did) plus a lot more is now implemented in IQ-TREE2. 

Rob

[shuiyin liu]

Hi Rob,

I met the similar problem with Teagan when I ran partitionfinder2 with two big datasets of 977 and 2821 genes. After over 20 days, it was still spinning at "Using a separate GTR+G model for each data block". Here is my command "python PartitionFinder.py RNA_977MOr2 --raxml --rcluster-max 100 -p 80", following the partitionfinder manual for big dataset. I noticed your above response with iqtree2 as alternative analytical method. But the command "iqtree -s example.phy -p example.nex -m MFP+MERGE -rcluster 100" is the same as the greedy algorithm. Is there a good way to run partitionfinder in iqtree2 for big dataset, such as the rcluster algorithm with rcluster-max=100?

Best wishes

Shuiyin

[shuiyin liu]

Hi Rob,

I met the similar problem with Teagan when I ran partitionfinder2 with two big datasets of 977 and 2821 genes. After over 20 days, it was still spinning at "Using a separate GTR+G model for each data block". Here is my command "python PartitionFinder.py RNA_977MOr2 --raxml --rcluster-max 100 -p 80", following the partitionfinder manual for big dataset. I noticed your above response with iqtree2 as alternative analytical method. But the command "iqtree -s example.phy -p example.nex -m MFP+MERGE -rcluster 100" is the same as the greedy algorithm. Is there a good way to run partitionfinder in iqtree2 for big dataset, such as the rcluster algorithm with rcluster-max=100?

Best wishes

Shuiyin

[shuiyin liu]

Hi Rob,

I met the same problem with Teagan when I ran partitionfinder2 with two big datasets of 977  and 2821 genes. After over 20-days, it was still spinning at "Using a separate GTR+G model for each data block" for these two datasets. Here is my command "python PartitionFinder.py RNA_977MOr2 --raxml --rcluster-max 100 -p 80", following the partitionfinder manual for big dataset. I noticed your above response with iqtree2 as alternative analytical method. But this command "iqtree -s example.phy -p example.nex -m MFP+MERGE -rcluster 100" is the same as the greedy algorithm. Is there a good way to run partitionfinder in iqtree2 for big dataset, such as using the rcluster algorithm and setting rcluster-max=100? 

Best wishe

Shuiyin