pyabpoa provides an easy-to-use interface to abPOA, it contains all the APIs that can be used to perform MSA for a set of sequences and consensus calling from the final alignment graph.

pyabpoa can be installed with pip:

pip install pyabpoa

Alternatively, you can install pyabpoa from source (cython is required):

git clone --recursive https://github.com/yangao07/abPOA.git
cd abPOA
make install_py

The following code illustrates how to use pyabpoa.

import pyabpoa as pa
a = pa.msa_aligner()
seqs=[
'CCGAAGA',
'CCGAACTCGA',
'CCCGGAAGA',
'CCGAAGA'
]
a_res=a.msa(seqs, out_cons=True, out_msa=True) # perform multiple sequence alignment 

for seq in a_res.cons_seq:
    print(seq)  # print consensus sequence

a_res.print_msa() # print row-column multiple sequence alignment in PIR format

# incrementally add new seqs
new_seqs=[
'CCAGA',
'CCGAAAGA'
]
b = pa.msa_aligner()
b.msa_align(seqs, out_cons=True, out_msa=True)
b.msa_add(new_seqs)
b_res = b.msa_output()
b_res.print_msa()

You can also try the example script provided in the source folder:

python ./python/example.py

pyabpoa.msa_aligner(aln_mode='g', ...)

This constructs a multiple sequence alignment handler of pyabpoa, it accepts the following arguments:

• aln_mode: alignment mode. 'g': global, 'l': local, 'e': extension; default: 'g'
• is_aa: input is amino acid sequence; default: False
• match: match score; default: 2
• mismatch: match penaty; default: 4
• score_matrix: scoring matrix file, match and mismatch are not used when score_matrix is used; default: ''
• gap_open1: first gap opening penalty; default: 4
• gap_ext1: first gap extension penalty; default: 2
• gap_open2: second gap opening penalty; default: 24
• gap_ext2: second gap extension penalty; default: 1
• extra_b: first adaptive banding paremeter; set as < 0 to disable adaptive banded DP; default: 10
• extra_f: second adaptive banding paremete; the number of extra bases added on both sites of the band is b+f*L, where L is the length of the aligned sequence; default : 0.01
• cons_algrm: consensus calling algorithm. 'HB': heaviest bunlding, 'MF': most frequent bases; default: 'HB'

The msa_aligner handler provides one method msa which performs multiple sequence alignment and accepts the following arguments:

pyabpoa.msa_aligner.msa(seqs, out_cons, out_msa, out_pog='', incr_fn='', qscores=None)

• seqs: a list variable containing a set of input sequences; positional
• out_cons: a bool variable to ask pyabpoa to generate consensus sequence; positional
• out_msa: a bool variable to ask pyabpoa to generate RC-MSA; positional
• max_n_cons: maximum number of consensus sequence to generate; default: 1
• min_freq: minimum frequency of each consensus to output (effective when max_n_cons > 1); default: 0.3
• out_pog: name of a file (.png or .pdf) to store the plot of the final alignment graph; default: ''
• incr_fn: name of an existing graph (GFA) or MSA (FASTA) file, incrementally align sequence to this graph/MSA; default: ''
• qscores: optional per-sequence quality information used to weight the consensus graph. Each entry must match the corresponding sequence length and should be a list of integer Phred scores, e.g. record.letter_annotations["phred_quality"] from Biopython; default: None

When qscores is provided, pyabpoa enables quality-weighted consensus (use_qv) for that run. This is effective for heaviest-bundling consensus (cons_algrm='HB'), matching the CLI -Q/--use-qual-weight behavior.

msa_aligner also provides three methods for incrementally adding sequences to graph/MSA:

pyabpoa.msa_aligner.msa_align(seqs, out_cons, out_msa, max_n_cons=1, min_freq=0.25, incr_fn=b'', qscores=None)
pyabpoa.msa_aligner.msa_add(new_seqs, qscores=None)
pyabpoa.msa_aligner.msa_output()

Intuitively, msa() = msa_align()+msa_add()+msa_output().

To collect consenus sequence and RC-MSA result, msa_output() needs to be called after msa_align() and msa_add(), which returns an object of pyabpoa.msa_result.

pyabpoa.msa_result(seq_n, cons_n, cons_len, ...)

This class describes the information of the generated consensus sequence and the RC-MSA. The returned result of pyabpoa.msa_aligner.msa() is an object of this class that has the following properties:

• n_seq: number of input aligned sequences
• n_cons: number of generated consensus sequences (generally 1, could be 2 or more if max_n_cons is set as > 1)
• clu_n_seq: an array of sequence cluster size
• cons_len: an array of consensus sequence length(s)
• cons_seq: an array of consensus sequence(s)
• cons_cov: an array of consensus sequence coverage for each base
• cons_qv: an array of consensus quality strings in FASTQ encoding
• msa_len: size of each row in the RC-MSA
• msa_seq: an array containing n_seq+n_cons strings that demonstrates the RC-MSA, each consisting of one input sequence and several - indicating the alignment gaps.

pyabpoa.msa_result() has a function of print_msa which prints the RC-MSA to screen.

pyabpoa.msa_result().print_msa()