fewer clones in cluster_assignment than clone_freq

[jamelee]

Hi,

@amcpherson Thans for developing this tool!

In my result, I got 6 clones in /trees/{tree_solution}/clone_freq, but only got 5 clones in /trees/{tree_solution}/cluster_assignment/.

The content of clone_freq is:

0.618809         0.0            0.0             0.0173867       0.0970734       0.266731
0.0484576       0.0809986       0.104325        0.112758        0.142484         0.510976

So there are 6 clones. Am I right?

The content of cluster_assignment is:

index      value
0            5
1            4
2            3
3            2
4            1
5            4

My guess is that the numbers in "value" are clone IDs. If it's the case, there are only 5 clones which is fewer than 6 as indicated by clone_freq.

Could you please help me out?

Thank you in advance.

[zhanyinx]

Hello,

I have a related problem.
I have 3 timepoints and the mutations have been clustered using pyclone (6 clusters (0 to 5))
I have run run_citup_qip.sh and got the following output:

The output of pd.read_hdf("out/results.h5","/trees/50/clone_freq")

        0         1       2         3       4       5       6
0  0.0121  0.008916  0.0000  0.604584  0.1734  0.1898  0.0112
1  0.0131  0.191100  0.4791  0.092900  0.0202  0.1940  0.0096
2  0.0148  0.132900  0.5346  0.031900  0.0101  0.2277  0.0480

Which shows 7 clones.

The cluster assignment output is:

0    4
1    6
2    3
3    5
4    2
5    1

I guess the first column is the cluster id input while the second is the clone. However, in the clone assignment, there is no clone 0. I was wondering how do I know which one is clone 0.

A wider question is: how do I know which mutation belongs to which clone?

Thanks
Best
Zhan

[jamelee]

According to the description in paper, I think clone 0 means healthy cells:
Note that while CITUP does not assign mutations to the root nodes in CLL003 and CLL077, the root node in CLL006 is assigned five mutations. This is in agreement with the observation in Schuh et al. (2012) that the normal contamination in this patient is insignificant and suggests that CITUP is able to automatically handle presence or absence of healthy cell contamination.

As for the question: how do I know which mutation belongs to which clone?
You know which cluster a mutation belongs to, which is input requried by citup, and now you have the relation between cluster ID and clone ID from cluster_assignment output, so it's easy to know which clone a mutation belongs to.