comparison blast2html.py @ 120:2729c2326235

Fix for Rikilt issue 13 Hit e-value and identity% should be taken from the hsp with the highest bit score. Previously each of these values was calculated independently. Also use arrays for cover calculation instead of python lists and refactor the hit_info() code a bit.

119:591dc9c24824

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

        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 = "{0:.1f}".format(max(hsp_val('Hsp_bit-score'))),
                       totalscore = "{0:.1f}".format(sum(hsp_val('Hsp_bit-score'))),
                       cover = "{0:.0%}".format(cover_count / query_length),
                       e_value = "{0:.4}".format(float(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, text_from='hitid', cssclass=None, display_nolink=True):
        """Create a html link from a hit node to a configured gene bank webpage.
        text: The text of the link. If not set applies text_from.

120:2729c2326235

import six, codecs, io
from six.moves import builtins
from os import path
from itertools import repeat
from collections import defaultdict, namedtuple
from array import array
import glob
import argparse
from lxml import objectify
import jinja2

        query_length = blastxml_len(result)

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

            cover = array('B', [0]) * query_length
            for hsp in hsps:
                cover[hsp['Hsp_query-from']-1 : int(hsp['Hsp_query-to'])] = array('B', [1]) * blastxml_len(hsp)
            cover_count = cover.count(1)

            best_hsp = max(hsps, key=lambda h: h['Hsp_bit-score'])

            yield dict(hit = hit,
                       title = firsttitle(hit),
                       maxscore = format(float(best_hsp['Hsp_bit-score']), '.1f'),
                       e_value = format(float(best_hsp.Hsp_evalue), '.4'),
                       # float(...) because non-flooring division doesn't work with lxml elements in python 2.6
                       ident = format(float(best_hsp.Hsp_identity) / blastxml_len(best_hsp), '.0%'),
                       totalscore = format(sum(hsp['Hsp_bit-score'] for hsp in hsps), '.1f'),
                       cover = format(cover_count / query_length, '.0%'),
                  )

    @filter
    def genelink(self, hit, text=None, text_from='hitid', cssclass=None, display_nolink=True):
        """Create a html link from a hit node to a configured gene bank webpage.
        text: The text of the link. If not set applies text_from.