comparison blast2html.py @ 98:4378d11f0ed7 draft

implement configurable gene bank links

97:df9fd5f35967

import warnings
import six, codecs
from six.moves import builtins
from os import path
from itertools import repeat
import argparse
from lxml import objectify
import jinja2



_filters = dict(float='float')
def filter(func_or_name):

    """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"""

        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, six.string_types):
        hit = hitid(hit)
    link = "http://www.ncbi.nlm.nih.gov/nucleotide/{0}?report={1}&log$=nuclalign".format(hit, type)
    if hsp != None:
        link += "&from={0}&to={1}".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.

    Javascript string literal escape. Note that this only escapes data
    for embedding within javascript string literals, not in general
    javascript snippets.
    """

    value = six.text_type(value)

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

    return value

    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)

        )

        result = template.render(blast=self.blast,
                                 iterations=self.blast.BlastOutput_iterations.Iteration,
                                 colors=self.colors,
                                 genelink=genelink,
                                 params=params)
        if six.PY2:
            result = result.encode('utf-8')
        output.write(result)

                       totalscore = "{0:.1f}".format(sum(hsp_val('Hsp_bit-score'))),
                       cover = "{0:.0%}".format(cover_count / query_length),
                       e_value = "{0:.4g}".format(min(hsp_val('Hsp_evalue'))),
                       # FIXME: is this the correct formula vv?
                       # float(...) because non-flooring division doesn't work with lxml elements in python 2.6
                       ident = "{0:.0%}".format(float(min(float(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')

    # 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()

98:4378d11f0ed7

import warnings
import six, codecs
from six.moves import builtins
from os import path
from itertools import repeat
from collections import defaultdict
import argparse
from lxml import objectify
import jinja2

builtin_str = str
str = six.text_type



_filters = dict(float='float')
def filter(func_or_name):

    """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:]:
        id, title = t.split(' ', 1)
        titles.append(argparse.Namespace(Hit_id = id,
                                         Hit_def = title,
                                         Hit_accession = '',
                                         getroottree = hit.getroottree))
    return titles

@filter
def hitid(hit):
    return str(hit.Hit_id)


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

        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, six.string_types):
#         hit = hitid(hit)
#     link = "http://www.ncbi.nlm.nih.gov/nucleotide/{0}?report={1}&log$=nuclalign".format(hit, type)
#     if hsp != None:
#         link += "&from={0}&to={1}".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.

    Javascript string literal escape. Note that this only escapes data
    for embedding 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

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

    max_scale_labels = 10

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

        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)

        )

        result = template.render(blast=self.blast,
                                 iterations=self.blast.BlastOutput_iterations.Iteration,
                                 colors=self.colors,
                                 params=params)
        if six.PY2:
            result = result.encode('utf-8')
        output.write(result)

                       totalscore = "{0:.1f}".format(sum(hsp_val('Hsp_bit-score'))),
                       cover = "{0:.0%}".format(cover_count / query_length),
                       e_value = "{0:.4g}".format(min(hsp_val('Hsp_evalue'))),
                       # FIXME: is this the correct formula vv?
                       # float(...) because non-flooring division doesn't work with lxml elements in python 2.6
                       ident = "{0:.0%}".format(float(min(float(hsp.Hsp_identity) / blastxml_len(hsp) for hsp in hsps))))

    @filter
    def genelink(self, hit, text=None, clas=None, display_nolink=True):
        if text is None:
            text = hitid(hit)
        db = hit.getroottree().getroot().BlastOutput_db
        if isinstance(self.genelinks, six.string_types):
            template = self.genelinks
        else:
            template = self.genelinks.get(db)
        if template is None:
            return text if display_nolink else ''
        args = dict(id=hitid(hit).split('|'),
                    fullid=hitid(hit),
                    defline=str(hit.Hit_def).split('|'),
                    fulldefline=str(hit.Hit_def),
                    accession=str(hit.Hit_accession))
        try:
            link = template.format(**args)
        except Exception as e:
            warnings.warn('Error in formatting gene bank link {} with {}: {}'.format(template, args, e))
            return text if display_nolink else ''
        classattr = 'class="{}" '.format(jinja2.escape(clas)) if clas is not None else ''
        return jinja2.Markup("<a {}href=\"{}\">{}</a>".format(classattr, jinja2.escape(link), jinja2.escape(text)))


def read_genelinks(dir):
    links = {}
    for f in ('blastdb.loc', 'blastdb_p.loc', 'blastdb_d.loc'):
        try:
            f = open(path.join(dir, f))
            for l in f.readlines():
                if l.strip().startswith('#'):
                    continue
                line = l.split('\t')
                try:
                    links[line[2]] = line[3]
                except IndexError:
                    continue
            f.close()
        except OSError:
            continue
    if not links:
        warnings.warn("No gene bank link templates found")
    return links


def main():
    default_template = path.join(path.dirname(__file__), 'blast2html.html.jinja')

    # 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')
    
    dblink_group = parser.add_mutually_exclusive_group()
    dblink_group.add_argument('--genelink-template', default='http://www.ncbi.nlm.nih.gov/nucleotide/{accession}?report=genbank&log$=nuclalign',
                              help="""A link template to link hits to a gene bank webpage. The template string is a 
                              Python format string. It can contain the following replacement elements: {id[N]}, {fullid}, 
                              {defline[N]}, {fulldefline}, {accession}, where N is a number. id[N] and defline[N] will be 
                              replaced by the Nth element of the id or defline, where '|' is the field separator. 

                              The default is 'http://www.ncbi.nlm.nih.gov/nucleotide/{accession}?report=genbank&log$=nuclalign',
                              which is a link to the NCBI nucleotide database.""")
    
    dblink_group.add_argument('--db-config-dir',
                              help="""The directory where databases are configured in blastdb*.loc files. These files
                              are consulted for creating a gene bank link. The files should be tab-separated tables (with lines
                              starting with '#' ignored), where the third field of a line should be a database path and the fourth
                              a genebank link template conforming to the --genelink-template option syntax.

                              This option is incompatible with --genelink-template.""")
    
    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 = '.'

    if args.db_config_dir is None:
        genelinks = args.genelink_template
    elif not path.isdir(args.db_config_dir):
        parser.error('db-config-dir does not exist or is not a directory')
    else:
        genelinks = read_genelinks(args.db_config_dir)

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


if __name__ == '__main__':
    main()