fix: tighten & relax yaml parsing constraints

[tristan-f-r]

This checks for object prefixes before using eval. Instead, the parsing goes as:

• Is the param an int?
• Is the param a float?
• Is the param one of our desired evaluation objects? (If this causes too many problems, we can probably just flag np. to always be allowed.)
• Parse as a string.

This does make config parsing perhaps more rigid, but this has the side bonus of allowing singleton values.

(I chose not to prefer eval here, as I do not want to see how the python eval system and the already broad YAML format interact with each other. I still do an extra iterable check so the "can't index X by string" error appears without a recognizable stacktrace.)

This addresses two out of the three wants from #33.

[ntalluri]

Could you add test cases in test/config.py for this updated change?

[tristan-f-r]

Thanks! I forgot we had a dedicated test file for configs.

[ntalluri]

I was testing the different list options that could be used for the parameters using the config options below.

algorithms:
      - name: "pathlinker"
        params:
              include: true
              run1:
                  k: range(100,201,100)

      - name: "omicsintegrator1"
        params:
              include: true
              run1:
                  b: [5, 6]
                  w: np.linspace(0,5,2)
                  d: 10
                  dummy_mode: "file" # Or "terminals", "all", "others"

      - name: "omicsintegrator2"
        params:
              include: true
              run1:
                  b: 4
                  g: 0
              run2:
                  b: [2, 4]
                  g: [3]
              run3:
                  b: np.arange(5,6)
                  g: 4

      - name: "meo"
        params:
              include: true
              run1:
                  max_path_length: 3
                  local_search: "Yes"
                  rand_restarts: [10]
              run2:
                  max_path_length: [2, 3]
                  local_search: ["Yes", "No"]
                  rand_restarts: 10

      - name: "mincostflow"
        params:
              include: true
              run1:
                  flow: 1 # The flow must be an int
                  capacity: [1, 2]

      - name: "allpairs"
        params:
              include: true

      - name: "domino"
        params:
              include: true
              run1:
                  slice_threshold: 0.3
                  module_threshold: 0.05
              run2:
                  slice_threshold: [0.2]
                  module_threshold: [0.05]
              run3: 
                  slice_threshold: np.logspace(0, 1, num=3)
                  module_threshold: [0.05]

I keep getting this error:

TypeError in file /Users/nehatalluri/Desktop/research/spras/Snakefile, line 21:
Object of type int64 is not JSON serializable
File "/Users/nehatalluri/Desktop/research/spras/Snakefile", line 21, in
File "/Users/nehatalluri/Desktop/research/spras/spras/config.py", line 35, in init_global
File "/Users/nehatalluri/Desktop/research/spras/spras/config.py", line 111, in init
File "/Users/nehatalluri/Desktop/research/spras/spras/config.py", line 260, in process_config
File "/Users/nehatalluri/Desktop/research/spras/spras/util.py", line 25, in hash_params_sha1_base32
File "/Users/nehatalluri/anaconda3/envs/spras/lib/python3.11/json/init.py", line 238, in dumps
File "/Users/nehatalluri/anaconda3/envs/spras/lib/python3.11/json/encoder.py", line 200, in encode
File "/Users/nehatalluri/anaconda3/envs/spras/lib/python3.11/json/encoder.py", line 258, in iterencode
File "/Users/nehatalluri/anaconda3/envs/spras/lib/python3.11/json/encoder.py", line 180, in default

The parsing stops after reading the omicsintegrator2 params, but I don't see any progress for reading MEO.

[tristan-f-r]

Interesting - I'll add more tests to cover all of these cases.

[tristan-f-r]

I do believe that that error is from the same origin as this "temporary" patch

spras/spras/config.py

Lines 246 to 247 in 73184ed

	if isinstance(value, np.float64):
	run_dict[param] = float(value)

[tristan-f-r]

It seems we never tested for np.arange, because this error is present in master too. [I have found the fix for this, I'm currently adding a test case.]

[tristan-f-r]

Adding more tests for some more code coverage.

[ntalluri]

Note: test egfr config