annotate blast2html.py @ 71:371cd585e459

refactor

#!/usr/bin/env python3
# -*- coding: utf-8 -*-

# Copyright The Hyve B.V. 2014
# License: GPL version 3 or higher

from __future__ import unicode_literals

import sys
import math
import warnings
from os import path
from itertools import repeat
import argparse
from lxml import objectify
import jinja2



_filters = {}
def filter(func_or_name):
    "Decorator to register a function as filter in the current jinja environment"
    if isinstance(func_or_name, str):
        def inner(func):
            _filters[func_or_name] = func.__name__
            return func
        return inner
    else:
        _filters[func_or_name.__name__] = func_or_name.__name__
        return func_or_name


def color_idx(length):
    if length < 40:
        return 0
    elif length < 50:
        return 1
    elif length < 80:
        return 2
    elif length < 200:
        return 3
    return 4

@filter
def fmt(val, fmt):
    return format(float(val), fmt)

@filter
def firsttitle(hit):
    return hit.Hit_def.text.split('>')[0]

@filter
def othertitles(hit):
    """Split a hit.Hit_def that contains multiple titles up, splitting out the hit ids from the titles."""
    id_titles = hit.Hit_def.text.split('>')

    titles = []
    for t in id_titles[1:]:
        fullid, title = t.split(' ', 1)
        hitid, id = fullid.split('|', 2)[1:3]
        titles.append(dict(id = id,
                           hitid = hitid,
                           fullid = fullid,
                           title = title))
    return titles

@filter
def hitid(hit):
    hitid = hit.Hit_id.text
    s = hitid.split('|', 2)
    if len(s) >= 2:
        return s[1]
    return hitid

@filter
def seqid(hit):
    hitid = hit.Hit_id.text
    s = hitid.split('|', 2)
    if len(s) >= 3:
        return s[2]
    return hitid


@filter
def alignment_pre(hsp):
    """Create the preformatted alignment blocks"""

    # line break length
    linewidth = 60

    qfrom = int(hsp['Hsp_query-from'])
    qto = int(hsp['Hsp_query-to'])
    qframe = int(hsp['Hsp_query-frame'])
    hfrom = int(hsp['Hsp_hit-from'])
    hto = int(hsp['Hsp_hit-to'])
    hframe = int(hsp['Hsp_hit-frame'])
    
    qseq = hsp.Hsp_qseq.text
    midline = hsp.Hsp_midline.text
    hseq = hsp.Hsp_hseq.text

    if not qframe in (1, -1):
        warnings.warn("Error in BlastXML input: Hsp node {} has a Hsp_query-frame of {}. (should be 1 or -1)".format(nodeid(hsp), qframe))
        qframe = -1 if qframe < 0 else 1
    if not hframe in (1, -1):
        warnings.warn("Error in BlastXML input: Hsp node {} has a Hsp_hit-frame of {}. (should be 1 or -1)".format(nodeid(hsp), hframe))
        hframe = -1 if hframe < 0 else 1
    
    def split(txt):
        return [txt[i:i+linewidth] for i in range(0, len(txt), linewidth)]

    for qs, mid, hs, offset in zip(split(qseq), split(midline), split(hseq), range(0, len(qseq), linewidth)):
        yield (
            "Query  {:>7}  {}  {}\n".format(qfrom+offset*qframe, qs, qfrom+(offset+len(qs)-1)*qframe) +
            "       {:7}  {}\n".format('', mid) +
            "Subject{:>7}  {}  {}".format(hfrom+offset*hframe, hs, hfrom+(offset+len(hs)-1)*hframe)
        )
        
    if qfrom+(len(qseq)-1)*qframe != qto:
        warnings.warn("Error in BlastXML input: Hsp node {} qseq length mismatch: from {} to {} length {}".format(
            nodeid(hsp), qfrom, qto, len(qseq)))
    if hfrom+(len(hseq)-1)*hframe != hto:
        warnings.warn("Error in BlastXML input: Hsp node {} hseq length mismatch: from {} to {} length {}".format(
            nodeid(hsp), hfrom, hto, len(hseq)))

    

@filter('len')
def blastxml_len(node):
    if node.tag == 'Hsp':
        return int(node['Hsp_align-len'])
    elif node.tag == 'Iteration':
        return int(node['Iteration_query-len'])
    raise Exception("Unknown XML node type: "+node.tag)

@filter
def nodeid(node):
    id = []
    if node.tag == 'Hsp':
        id.insert(0, node.Hsp_num.text)
        node = node.getparent().getparent()
        assert node.tag == 'Hit'
    if node.tag == 'Hit':
        id.insert(0, node.Hit_num.text)
        node = node.getparent().getparent()
        assert node.tag == 'Iteration'
    if node.tag == 'Iteration':
        id.insert(0, node['Iteration_iter-num'].text)
        return '-'.join(id)
    raise ValueError("The nodeid filter can only be applied to Hsp, Hit or Iteration nodes in a BlastXML document")

    
@filter
def asframe(frame):
    if frame == 1:
        return 'Plus'
    elif frame == -1:
        return 'Minus'
    raise Exception("frame should be either +1 or -1")

def genelink(hit, type='genbank', hsp=None):
    if not isinstance(hit, str):
        hit = hitid(hit)
    link = "http://www.ncbi.nlm.nih.gov/nucleotide/{}?report={}&log$=nuclalign".format(hit, type)
    if hsp != None:
        link += "&from={}&to={}".format(hsp['Hsp_hit-from'], hsp['Hsp_hit-to'])
    return link


# javascript escape filter based on Django's, from https://github.com/dsissitka/khan-website/blob/master/templatefilters.py#L112-139
# I've removed the html escapes, since html escaping is already being performed by the template engine.

_base_js_escapes = (
    ('\\', r'\u005C'),
    ('\'', r'\u0027'),
    ('"', r'\u0022'),
    # ('>', r'\u003E'),
    # ('<', r'\u003C'),
    # ('&', r'\u0026'),
    # ('=', r'\u003D'),
    # ('-', r'\u002D'),
    # (';', r'\u003B'),
    # (u'\u2028', r'\u2028'),
    # (u'\u2029', r'\u2029')
)

# Escape every ASCII character with a value less than 32. This is
# needed a.o. to prevent html parsers from jumping out of javascript
# parsing mode.
_js_escapes = (_base_js_escapes +
               tuple(('%c' % z, '\\u%04X' % z) for z in range(32)))

@filter
def js_string_escape(value):
    """Escape javascript string literal escapes. Note that this only works
    within javascript string literals, not in general javascript
    snippets."""

    value = str(value)

    for bad, good in _js_escapes:
        value = value.replace(bad, good)

    return value

@filter
def hits(result):
    # sort hits by longest hotspot first
    return sorted(result.Iteration_hits.findall('Hit'),
                  key=lambda h: max(blastxml_len(hsp) for hsp in h.Hit_hsps.Hsp),
                  reverse=True)



class BlastVisualize:

    colors = ('black', 'blue', 'green', 'magenta', 'red')

    max_scale_labels = 10

    def __init__(self, input, templatedir, templatename):
        self.input = input
        self.templatename = templatename

        self.blast = objectify.parse(self.input).getroot()
        self.loader = jinja2.FileSystemLoader(searchpath=templatedir)
        self.environment = jinja2.Environment(loader=self.loader,
                                              lstrip_blocks=True, trim_blocks=True, autoescape=True)

        self._addfilters(self.environment)


    def _addfilters(self, environment):
        for filtername, funcname in _filters.items():
            try:
                environment.filters[filtername] = getattr(self, funcname)
            except AttributeError:
                environment.filters[filtername] = globals()[funcname]

    def render(self, output):
        template = self.environment.get_template(self.templatename)

        params = (('Query ID', self.blast["BlastOutput_query-ID"]),
                  ('Query definition', self.blast["BlastOutput_query-def"]),
                  ('Query length', self.blast["BlastOutput_query-len"]),
                  ('Program', self.blast.BlastOutput_version),
                  ('Database', self.blast.BlastOutput_db),
        )

        output.write(template.render(blast=self.blast,
                                     iterations=self.blast.BlastOutput_iterations.Iteration,
                                     colors=self.colors,
                                     # match_colors=self.match_colors(),
                                     # hit_info=self.hit_info(),
                                     genelink=genelink,
                                     params=params))

    @filter
    def match_colors(self, result):
        """
        An iterator that yields lists of length-color pairs. 
        """

        query_length = blastxml_len(result)
        
        percent_multiplier = 100 / query_length

        for hit in hits(result):
            # sort hotspots from short to long, so we can overwrite index colors of
            # short matches with those of long ones.
            hotspots = sorted(hit.Hit_hsps.Hsp, key=lambda hsp: blastxml_len(hsp))
            table = bytearray([255]) * query_length
            for hsp in hotspots:
                frm = hsp['Hsp_query-from'] - 1
                to = int(hsp['Hsp_query-to'])
                table[frm:to] = repeat(color_idx(blastxml_len(hsp)), to - frm)

            matches = []
            last = table[0]
            count = 0
            for i in range(query_length):
                if table[i] == last:
                    count += 1
                    continue
                matches.append((count * percent_multiplier, self.colors[last] if last != 255 else 'transparent'))
                last = table[i]
                count = 1
            matches.append((count * percent_multiplier, self.colors[last] if last != 255 else 'transparent'))

            yield dict(colors=matches, hit=hit, defline=firsttitle(hit))

    @filter
    def queryscale(self, result):
        query_length = blastxml_len(result)
        skip = math.ceil(query_length / self.max_scale_labels)
        percent_multiplier = 100 / query_length
        for i in range(1, query_length+1):
            if i % skip == 0:
                yield dict(label = i, width = skip * percent_multiplier)
        if query_length % skip != 0:
            yield dict(label = query_length,
                       width = (query_length % skip) * percent_multiplier)

    @filter
    def hit_info(self, result):

        query_length = blastxml_len(result)

        for hit in hits(result):
            hsps = hit.Hit_hsps.Hsp

            cover = [False] * query_length
            for hsp in hsps:
                cover[hsp['Hsp_query-from']-1 : int(hsp['Hsp_query-to'])] = repeat(True, blastxml_len(hsp))
            cover_count = cover.count(True)

            def hsp_val(path):
                return (float(hsp[path]) for hsp in hsps)

            yield dict(hit = hit,
                      title = firsttitle(hit),
                       maxscore = "{:.1f}".format(max(hsp_val('Hsp_bit-score'))),
                       totalscore = "{:.1f}".format(sum(hsp_val('Hsp_bit-score'))),
                       cover = "{:.0%}".format(cover_count / query_length),
                       e_value = "{:.4g}".format(min(hsp_val('Hsp_evalue'))),
                       # FIXME: is this the correct formula vv?
                       ident = "{:.0%}".format(float(min(hsp.Hsp_identity / blastxml_len(hsp) for hsp in hsps))),
                       accession = hit.Hit_accession)


def main():
    default_template = path.join(path.dirname(__file__), 'blast2html.html.jinja')
    
    parser = argparse.ArgumentParser(description="Convert a BLAST XML result into a nicely readable html page",
                                     usage="{} [-i] INPUT [-o OUTPUT]".format(sys.argv[0]))
    input_group = parser.add_mutually_exclusive_group(required=True)
    input_group.add_argument('positional_arg', metavar='INPUT', nargs='?', type=argparse.FileType(mode='r'),
                             help='The input Blast XML file, same as -i/--input')
    input_group.add_argument('-i', '--input', type=argparse.FileType(mode='r'), 
                             help='The input Blast XML file')
    parser.add_argument('-o', '--output', type=argparse.FileType(mode='w'), default=sys.stdout,
                        help='The output html file')
    # We just want the file name here, so jinja can open the file
    # itself. But it is easier to just use a FileType so argparse can
    # handle the errors. This introduces a small race condition when
    # jinja later tries to re-open the template file, but we don't
    # care too much.
    parser.add_argument('--template', type=argparse.FileType(mode='r'), default=default_template,
                        help='The template file to use. Defaults to blast_html.html.jinja')

    args = parser.parse_args()
    if args.input == None:
        args.input = args.positional_arg
    if args.input == None:
        parser.error('no input specified')

    templatedir, templatename = path.split(args.template.name)
    args.template.close()
    if not templatedir:
        templatedir = '.'

    b = BlastVisualize(args.input, templatedir, templatename)
    b.render(args.output)


if __name__ == '__main__':
    main()