/SeqAlign

Primary LanguagePythonMIT LicenseMIT

SeqAlign: Sequence Alignment

Sequence alignment done right -- Modules implementing sequence alignment algorithms correctly.. 🤓. See the paper by Flouri et al. for further discussion about the correctness of sequence alignment algorithms.

Moreover, the implemented alignment algorithms not only operate on sequences composed of character elements generated from a given alphabet, but also on generic sequence elements (i.e. nodes) where each element/node has some feature vector representation. We only need to define/implement a similarity measure/function that computes the similarity between the elements/nodes in the sequences being aligned.

Installation

  • git clone the repo and cd into it.
  • Run pip install -e . to install the repo's python package.

TODO

  1. Provide examples for non character elements in sequence alignment
  2. Provide examples for character sequence alignment (for now check the tests module under src folder)

Non-temporal Aligner

Supported alignment methods

  1. global alignment (i.e. Needlman-Wunch or more precisely a variant of David Sankoff algorithm). See Sankoff, D. (1972). "Matching sequences under deletion-insertion constraints". Proceedings of the National Academy of Sciences of the United States of America. 69 (1): 4–6
  2. local alignment (i.e. Smith-Waterman algorithm)
  3. semi-global alignment (shorter to longer sequences)
  4. end-gap-free alignment (i.e. detecting overlap)

Supported alignment penalties

The current implementation allows for specifying:

  • linear gap penalty
  • affine gap penalty

Temporal Aligner

Based on the work reported in Syed et al. Temporal Needleman-Wunsch and further expanded to support other alignment methods and penalties (see below).

Supported alignment methods

  1. global alignment (i.e. Needlman-Wunch or more precisely a variant of David Sankoff algorithm). See Sankoff, D. (1972). "Matching sequences under deletion-insertion constraints". Proceedings of the National Academy of Sciences of the United States of America. 69 (1): 4–6
  2. local alignment (i.e. Smith-Waterman algorithm)
  3. semi-global alignment (shorter to longer sequences)
  4. end-gap-free alignment (i.e. detecting overlap)

Supported alignment penalties

The current implementation allows for specifying:

  • linear gap penalty

License

MIT License file accompanies this repository.

NB: All implemented algorithms return alignment score with all the possible alignment paths having the optimal score.