search

Roth-Lab / pyclone

Probabilistic model for inferring clonal population structure from deep NGS sequencing.
https://bitbucket.org/aroth85/pyclone/wiki/Home
Other
98 stars 36 forks source link

Fail to reproduce tutorial sample results in 2014 paper #7

Open gengyuncong opened 5 years ago

gengyuncong commented 5 years ago

Hi,

I analyze the tutorial data provided using the latest PyClone version. However, I ended up with 9 clusters which is different from results of the nature paper published in 2014. I have some questions:

  1. Are the tutorial data (SRR385938-SRR385941) same as the one used in nature paper?
  2. Do I need to tune any parameters in order to reproduce the result?
  3. Is there any inconsistency due to version update? Thank you!

Here is what I got: test.pdf

Here is my config file:

num_iters: 10000

base_measure_params:
 alpha: 1
 beta: 1

concentration:
 value: 1.0

 prior:
   shape: 1.0
   rate: 0.001

density: pyclone_beta_binomial

beta_binomial_precision_params:
 value: 1000

 prior:
   shape: 1.0
   rate: 0.0001

 proposal:
   precision: 0.01

working_dir: /Users/apple/FCBB-final/data/tutorial

trace_dir: trace

samples:
 SRR385938:
   mutations_file: yaml/SRR385938.yaml

   tumour_content:
     value: 1.0

   error_rate: 0.001

 SRR385939:
   mutations_file: yaml/SRR385939.yaml

   tumour_content:
     value: 1.0

   error_rate: 0.001

 SRR385940:
   mutations_file: yaml/SRR385940.yaml

   tumour_content:
     value: 1.0

   error_rate: 0.001

 SRR385941:
   mutations_file: yaml/SRR385941.yaml

   tumour_content:
     value: 1.0

   error_rate: 0.001
gengyuncong commented 5 years ago

And could you provide the ids of the mutation which are (i) shared by NA18507 and NA19240, (ii) shared by NA18507, NA19240 and NA12878 and (iii) shared by all four cases?