--fit-model error

[luciagrami]

Hello,
I am using poppunk 2.7.0 with pp-sketchlib v2.1.4.

I am having issues when running fit model:

poppunk --fit-model dbscan --ref-db TBdb --output TBdb_hdbscan

Output:
PopPUNK (POPulation Partitioning Using Nucleotide Kmers)
(with backend: sketchlib v2.1.4
sketchlib: /atlas/apps/miniconda3/envs/pp_env/lib/python3.11/site-packages/pp_sketchlib.cpython-311-x86_64-linux-gnu.so)

Graph-tools OpenMP parallelisation enabled: with 35 threads
Mode: Fitting dbscan model to reference database

Fitting HDBSCAN model using a CPU
Fitting HDBSCAN model using a CPU
Fitting HDBSCAN model using a CPU
Fitting HDBSCAN model using a CPU
Assigning distances with DBSCAN model
100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 1026/1026 [00:26<00:00, 38.10it/s]
Fit summary:
Number of clusters 21
Number of datapoints 100000
Number of assignments 81096

Scaled component means
[0.65383488 0.22816747]
[0.38883853 0.14202529]
[0.09366661 0.00975697]
[2.56513596e-01 4.57124497e-06]
[0.22594947 0. ]
[4.91949245e-02 4.69420002e-06]
[7.73994699e-02 2.12680433e-07]
[1.55685291e-01 3.80115409e-07]
[8.13707039e-02 9.56804570e-07]
[0.08370336 0. ]
[8.88103917e-02 5.63386891e-07]
[0.0864341 0. ]
[0.12166263 0. ]
[1.23963781e-01 1.00648538e-06]
[0.09071088 0. ]
[0.09283514 0. ]
[1.10260807e-01 9.44051010e-07]
[1.17498189e-01 5.60512490e-08]
[0.0949481 0. ]
[9.98577848e-02 8.35967001e-07]
[1.04917660e-01 6.65403661e-07]

Network summary:
Components 1547
Density 0.0163
Transitivity 0.6403
Mean betweenness 0.3620
Weighted-mean betweenness 0.1349
Score 0.6299
Score (w/ betweenness) 0.4019
Score (w/ weighted-betweenness) 0.5449
Traceback (most recent call last):
File "/atlas/apps/miniconda3/envs/pp_env/bin/poppunk", line 11, in
sys.exit(main())
^^^^^^
File "/atlas/apps/miniconda3/envs/pp_env/lib/python3.11/site-packages/PopPUNK/main.py", line 668, in main
isolateClustering = {fit_type: printClusters(genomeNetwork,
^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/atlas/apps/miniconda3/envs/pp_env/lib/python3.11/site-packages/PopPUNK/network.py", line 1520, in printClusters
unword = next(unword_generator)
^^^^^^^^^^^^^^^^^^^^^^
File "/atlas/apps/miniconda3/envs/pp_env/lib/python3.11/site-packages/PopPUNK/unwords.py", line 31, in gen_unword
word += "".join(syllable()0)
^^^^^^^^^^^^^^^
File "/atlas/apps/miniconda3/envs/pp_env/lib/python3.11/site-packages/PopPUNK/unwords.py", line 20, in
cv = lambda: consonant() + vowel()
^^^^^^^^^^^
File "/atlas/apps/miniconda3/envs/pp_env/lib/python3.11/site-packages/PopPUNK/unwords.py", line 19, in
consonant = lambda: random.sample(consonants, 1)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/atlas/apps/miniconda3/envs/pp_env/lib/python3.11/random.py", line 439, in sample
raise TypeError("Population must be a sequence. "
TypeError: Population must be a sequence. For dicts or sets, use sorted(d).

The error occurs when trying different models. Any suggestions? Thanks!

[johnlees]

Hopefully a fix for this in #322, which should appear in v2.7.1.

I'm not sure why that has suddenly started happening, and why it doesn't happen in the tests. Also makes it hard for me to verify this does fix the issue!

Can you change the following line in PopPUNK/unwords.py:
https://github.com/bacpop/PopPUNK/pull/322/files

To do this you'll need to clone the repository, and run poppunk with python poppunk-runner.py instead of just poppunk.

[johnlees]

i.e. on line 16 add sorted() around the right hand side
consonants = sorted(set(string.ascii_lowercase) - set(vowels) - set(trouble))

[johnlees]

Fixed in v2.7.1