SequenceFeature.get_split_kws

static SequenceFeature.get_split_kws(split_types=None, n_split_min=1, n_split_max=15, steps_pattern=None, n_min=2, n_max=4, len_max=15, steps_periodicpattern=None)[source]

Create dictionary with kwargs for three split types:

  • Segment: continuous sub-sequence.

  • Pattern: non-periodic discontinuous sub-sequence

  • PeriodicPattern: periodic discontinuous sub-sequence.

Added in version 0.1.0.

Parameters:

  • split_types (list of str, default=[Segment, Pattern, PeriodicPattern]) – Split types for which parameter dictionary should be generated.

  • n_split_min (int, default=1) – Number to specify the greatest Segment. Should be > 0.

  • n_split_max (int, default=15,) – Number to specify the smallest Segment. Should be >= n_split_min.

  • steps_pattern (list of int, default=[3, 4], optional) – Possible steps sizes for Pattern. Should contain at least 1 non-negative integers if Pattern split_type is used. If None, default is used.

  • n_min (int, default=2) – Minimum number of steps for Pattern. Should be <= n_max.

  • n_max (int, default=4) – Maximum number of steps for Pattern. Should be >= n_min.

  • len_max (int, default=15) – Maximum length in amino acid position for Pattern by varying start position. Should be > min(steps_pattern).

  • steps_periodicpattern (list of int, default=[3, 4], optional) – Size of odd and even steps for PeriodicPattern. Should contain two non-negative integers if PeriodicPattern split_type is used. If None, default is used.

Returns:

split_kws – Nested dictionary with parameters for chosen split_types:

  • Segment: {n_split_min:1, n_split_max=15}

  • Pattern: {steps=[3, 4], n_min=2, n_max=4, len_max=15}

  • PeriodicPattern: {steps=[3, 4]}

Return type:

dict

Notes

The split bounds returned here are validated for internal consistency (e.g., n_split_min <= n_split_max, n_min <= n_max, len_max > min(steps_pattern)); inconsistent values raise a ValueError.

Beyond that, the feasible maxima are effectively capped by the CPP part lengths used to build df_parts (tmd_len, jmd_n_len, jmd_c_len; defaults 20/10/10): a Segment cannot be split into more pieces than its part has residues, and a Pattern/PeriodicPattern cannot span beyond the part length. Choosing n_split_max, n_max, or len_max larger than a part can accommodate does not raise here — those splits simply yield empty feature buckets downstream. The one config that is always degenerate regardless of part length, an empty Pattern bucket where even the shortest repeat exceeds len_max (n_min * min(steps_pattern) > len_max), emits a UserWarning naming the offending parameters so it can be fixed by raising len_max or lowering steps_pattern/n_min.

Examples

Get default arguments for all splits types (Segment, Pattern, PeriodicPattern):

import aaanalysis as aa
sf = aa.SequenceFeature()
split_kws = sf.get_split_kws()
split_kws

{'Segment': {'n_split_min': 1, 'n_split_max': 15},
 'Pattern': {'steps': [3, 4], 'n_min': 2, 'n_max': 4, 'len_max': 15},
 'PeriodicPattern': {'steps': [3, 4]}}

You can also retrieve arguments for specific split types:

split_kws = sf.get_split_kws(split_types=["Segment", "Pattern"])
split_kws

{'Segment': {'n_split_min': 1, 'n_split_max': 15},
 'Pattern': {'steps': [3, 4], 'n_min': 2, 'n_max': 4, 'len_max': 15}}

The arguments for each split type can be adjusted. For Segments, their minimum and maximum length can be changed by the n_split_min (default=1) and n_split_max (default=15) parameters:

split_kws = sf.get_split_kws(split_types="Segment", n_split_min=5, n_split_max=10)
split_kws

{'Segment': {'n_split_min': 5, 'n_split_max': 10}}

For PeriodicPattern, the step size of each odd and even step can be specified as follows using the steps_periodicpattern (default=[3, 4]):

split_kws = sf.get_split_kws(split_types="PeriodicPattern", steps_periodicpattern=[5, 10])
split_kws

{'PeriodicPattern': {'steps': [5, 10]}}

And for Patterns, the step size, the minimum and maximum number of steps, and the maximum residue size of the pattern can be adjusted using the steps_pattern (default=[3, 4]), n_min (default=2), n_max (default=4), and len_max (default=10) parameters:

split_kws = sf.get_split_kws(split_types="Pattern", steps_pattern=[5, 10], n_min=3, n_max=5, len_max=30)
split_kws

{'Pattern': {'steps': [5, 10], 'n_min': 3, 'n_max': 5, 'len_max': 30}}