Пример #1
0
def run(args):
  import iotbx.phil
  from xia2.Modules.Scaler.rebatch import rebatch
  processed = iotbx.phil.process_command_line(args, master_phil)
  params = processed.work.extract()
  args = processed.remaining_args
  if params.hklin is None and len(args):
    params.hklin = args[0]
  assert params.hklin is not None

  rebatch(params.hklin, params.hklout,
          first_batch=params.first_batch,
          add_batch=params.add_batch,
          include_range=params.include_range,
          exclude_range=params.exclude_range)
Пример #2
0
def remove_blank(hklin, hklout):
    """Find and remove blank batches from the file. Returns hklin if no
    blanks."""

    blanks, goods = find_blank(hklin)

    if not blanks:
        return hklin

    # if mostly blank return hklin too...
    if len(blanks) > len(goods):
        logger.debug("%d blank vs. %d good: ignore", len(blanks), len(goods))
        return hklin

    rebatch(hklin, hklout, exclude_batches=blanks)

    return hklout
Пример #3
0
def remove_blank(hklin, hklout):
  '''Find and remove blank batches from the file. Returns hklin if no
  blanks.'''

  blanks, goods = find_blank(hklin)

  if not blanks:
    return hklin

  # if mostly blank return hklin too...
  if len(blanks) > len(goods):
    Debug.write('%d blank vs. %d good: ignore' % (len(blanks), len(goods)))
    return hklin

  from xia2.Modules.Scaler.rebatch import rebatch
  rebatch(hklin, hklout, exclude_batches=blanks)

  return hklout
Пример #4
0
    def _scale_prepare(self):
        """Perform all of the preparation required to deliver the scaled
        data. This should sort together the reflection files, ensure that
        they are correctly indexed (via pointless) and generally tidy
        things up."""

        # acknowledge all of the programs we are about to use...

        Citations.cite("pointless")
        Citations.cite("aimless")
        Citations.cite("ccp4")

        # ---------- GATHER ----------

        self._sweep_handler = SweepInformationHandler(self._scalr_integraters)

        for epoch in self._sweep_handler.get_epochs():
            si = self._sweep_handler.get_sweep_information(epoch)
            pname, xname, dname = si.get_project_info()
            sname = si.get_sweep_name()

            exclude_sweep = False

            for sweep in PhilIndex.params.xia2.settings.sweep:
                if sweep.id == sname and sweep.exclude:
                    exclude_sweep = True
                    break

            if exclude_sweep:
                self._sweep_handler.remove_epoch(epoch)
                logger.debug("Excluding sweep %s", sname)
            else:
                logger.debug("%-30s %s/%s/%s", "adding data from:", xname, dname, sname)

        # gather data for all images which belonged to the parent
        # crystal - allowing for the fact that things could go wrong
        # e.g. epoch information not available, exposure times not in
        # headers etc...

        for e in self._sweep_handler.get_epochs():
            si = self._sweep_handler.get_sweep_information(e)
            assert is_mtz_file(si.get_reflections()), repr(si.get_reflections())

        p, x = self._sweep_handler.get_project_info()
        self._scalr_pname = p
        self._scalr_xname = x

        # verify that the lattices are consistent, calling eliminate if
        # they are not N.B. there could be corner cases here

        need_to_return = False

        multi_sweep_indexing = PhilIndex.params.xia2.settings.multi_sweep_indexing

        # START OF if more than one epoch
        if len(self._sweep_handler.get_epochs()) > 1:

            # if we have multi-sweep-indexing going on then logic says all should
            # share common lattice & UB definition => this is not used here?

            # START OF if multi_sweep indexing and not input pg
            if multi_sweep_indexing and not self._scalr_input_pointgroup:
                pointless_hklins = []

                max_batches = 0
                for epoch in self._sweep_handler.get_epochs():
                    si = self._sweep_handler.get_sweep_information(epoch)
                    hklin = si.get_reflections()

                    batches = MtzUtils.batches_from_mtz(hklin)
                    if 1 + max(batches) - min(batches) > max_batches:
                        max_batches = max(batches) - min(batches) + 1

                logger.debug("Biggest sweep has %d batches", max_batches)
                max_batches = nifty_power_of_ten(max_batches)

                counter = 0

                refiners = []

                for epoch in self._sweep_handler.get_epochs():
                    si = self._sweep_handler.get_sweep_information(epoch)
                    hklin = si.get_reflections()
                    integrater = si.get_integrater()
                    refiner = integrater.get_integrater_refiner()
                    refiners.append(refiner)

                    hklin = self._prepare_pointless_hklin(
                        hklin, si.get_integrater().get_phi_width()
                    )

                    hklout = os.path.join(
                        self.get_working_directory(),
                        "%s_%s_%s_%s_prepointless.mtz"
                        % (pname, xname, dname, si.get_sweep_name()),
                    )

                    # we will want to delete this one exit
                    FileHandler.record_temporary_file(hklout)

                    first_batch = min(si.get_batches())
                    si.set_batch_offset(counter * max_batches - first_batch + 1)

                    rebatch(
                        hklin,
                        hklout,
                        first_batch=counter * max_batches + 1,
                        pname=pname,
                        xname=xname,
                        dname=dname,
                    )

                    pointless_hklins.append(hklout)

                    # update the counter & recycle
                    counter += 1

                    # SUMMARY - have added all sweeps to pointless_hklins

                s = self._factory.Sortmtz()

                pointless_hklin = os.path.join(
                    self.get_working_directory(),
                    "%s_%s_prepointless_sorted.mtz"
                    % (self._scalr_pname, self._scalr_xname),
                )

                s.set_hklout(pointless_hklin)

                for hklin in pointless_hklins:
                    s.add_hklin(hklin)

                s.sort()

                # FIXME xia2-51 in here look at running constant scaling on the
                # pointless hklin to put the runs on the same scale. Ref=[A]

                pointless_const = os.path.join(
                    self.get_working_directory(),
                    "%s_%s_prepointless_const.mtz"
                    % (self._scalr_pname, self._scalr_xname),
                )
                FileHandler.record_temporary_file(pointless_const)

                aimless_const = self._factory.Aimless()
                aimless_const.set_hklin(pointless_hklin)
                aimless_const.set_hklout(pointless_const)
                aimless_const.const()

                pointless_const = os.path.join(
                    self.get_working_directory(),
                    "%s_%s_prepointless_const_unmerged.mtz"
                    % (self._scalr_pname, self._scalr_xname),
                )
                FileHandler.record_temporary_file(pointless_const)
                pointless_hklin = pointless_const

                # FIXME xia2-51 in here need to pass all refiners to ensure that the
                # information is passed back to all of them not just the last one...
                logger.debug(
                    "Running multisweep pointless for %d sweeps", len(refiners)
                )
                pointgroup, reindex_op, ntr, pt = self._pointless_indexer_multisweep(
                    pointless_hklin, refiners
                )

                logger.debug("X1698: %s: %s", pointgroup, reindex_op)

                lattices = [Syminfo.get_lattice(pointgroup)]

                for epoch in self._sweep_handler.get_epochs():
                    si = self._sweep_handler.get_sweep_information(epoch)
                    intgr = si.get_integrater()
                    hklin = si.get_reflections()
                    refiner = intgr.get_integrater_refiner()

                    if ntr:
                        intgr.integrater_reset_reindex_operator()
                        need_to_return = True

                # SUMMARY - added all sweeps together into an mtz, ran
                # _pointless_indexer_multisweep on this, made a list of one lattice
                # and potentially reset reindex op?
            # END OF if multi_sweep indexing and not input pg

            # START OF if not multi_sweep, or input pg given
            else:
                lattices = []

                for epoch in self._sweep_handler.get_epochs():

                    si = self._sweep_handler.get_sweep_information(epoch)
                    intgr = si.get_integrater()
                    hklin = si.get_reflections()
                    refiner = intgr.get_integrater_refiner()

                    if self._scalr_input_pointgroup:
                        pointgroup = self._scalr_input_pointgroup
                        reindex_op = "h,k,l"
                        ntr = False

                    else:
                        pointless_hklin = self._prepare_pointless_hklin(
                            hklin, si.get_integrater().get_phi_width()
                        )

                        pointgroup, reindex_op, ntr, pt = self._pointless_indexer_jiffy(
                            pointless_hklin, refiner
                        )

                        logger.debug("X1698: %s: %s", pointgroup, reindex_op)

                    lattice = Syminfo.get_lattice(pointgroup)

                    if lattice not in lattices:
                        lattices.append(lattice)

                    if ntr:
                        intgr.integrater_reset_reindex_operator()
                        need_to_return = True
                # SUMMARY do pointless_indexer on each sweep, get lattices and make a list
                # of unique lattices, potentially reset reindex op.
            # END OF if not multi_sweep, or input pg given

            # SUMMARY - still within if more than one epoch, now have a list of number
            # of lattices

            # START OF if multiple-lattices
            if len(lattices) > 1:

                # why not using pointless indexer jiffy??!

                correct_lattice = sort_lattices(lattices)[0]

                logger.info("Correct lattice asserted to be %s", correct_lattice)

                # transfer this information back to the indexers
                for epoch in self._sweep_handler.get_epochs():

                    si = self._sweep_handler.get_sweep_information(epoch)
                    refiner = si.get_integrater().get_integrater_refiner()
                    sname = si.get_sweep_name()

                    state = refiner.set_refiner_asserted_lattice(correct_lattice)

                    if state == refiner.LATTICE_CORRECT:
                        logger.info(
                            "Lattice %s ok for sweep %s", correct_lattice, sname
                        )
                    elif state == refiner.LATTICE_IMPOSSIBLE:
                        raise RuntimeError(
                            f"Lattice {correct_lattice} impossible for {sname}"
                        )
                    elif state == refiner.LATTICE_POSSIBLE:
                        logger.info(
                            "Lattice %s assigned for sweep %s", correct_lattice, sname
                        )
                        need_to_return = True
            # END OF if multiple-lattices
            # SUMMARY - forced all lattices to be same and hope its okay.
        # END OF if more than one epoch

        # if one or more of them was not in the lowest lattice,
        # need to return here to allow reprocessing

        if need_to_return:
            self.set_scaler_done(False)
            self.set_scaler_prepare_done(False)
            return

        # ---------- REINDEX ALL DATA TO CORRECT POINTGROUP ----------

        # all should share the same pointgroup, unless twinned... in which
        # case force them to be...

        pointgroups = {}
        reindex_ops = {}
        probably_twinned = False

        need_to_return = False

        multi_sweep_indexing = PhilIndex.params.xia2.settings.multi_sweep_indexing

        # START OF if multi-sweep and not input pg
        if multi_sweep_indexing and not self._scalr_input_pointgroup:
            pointless_hklins = []

            max_batches = 0
            for epoch in self._sweep_handler.get_epochs():
                si = self._sweep_handler.get_sweep_information(epoch)
                hklin = si.get_reflections()

                batches = MtzUtils.batches_from_mtz(hklin)
                if 1 + max(batches) - min(batches) > max_batches:
                    max_batches = max(batches) - min(batches) + 1

            logger.debug("Biggest sweep has %d batches", max_batches)
            max_batches = nifty_power_of_ten(max_batches)

            counter = 0

            refiners = []

            for epoch in self._sweep_handler.get_epochs():
                si = self._sweep_handler.get_sweep_information(epoch)
                hklin = si.get_reflections()
                integrater = si.get_integrater()
                refiner = integrater.get_integrater_refiner()
                refiners.append(refiner)

                hklin = self._prepare_pointless_hklin(
                    hklin, si.get_integrater().get_phi_width()
                )

                hklout = os.path.join(
                    self.get_working_directory(),
                    "%s_%s_%s_%s_prepointless.mtz"
                    % (pname, xname, dname, si.get_sweep_name()),
                )

                # we will want to delete this one exit
                FileHandler.record_temporary_file(hklout)

                first_batch = min(si.get_batches())
                si.set_batch_offset(counter * max_batches - first_batch + 1)

                rebatch(
                    hklin,
                    hklout,
                    first_batch=counter * max_batches + 1,
                    pname=pname,
                    xname=xname,
                    dname=dname,
                )

                pointless_hklins.append(hklout)

                # update the counter & recycle
                counter += 1

            # FIXME related to xia2-51 - this looks very very similar to the logic
            # in [A] above - is this duplicated logic?
            s = self._factory.Sortmtz()

            pointless_hklin = os.path.join(
                self.get_working_directory(),
                "%s_%s_prepointless_sorted.mtz"
                % (self._scalr_pname, self._scalr_xname),
            )

            s.set_hklout(pointless_hklin)

            for hklin in pointless_hklins:
                s.add_hklin(hklin)

            s.sort()

            pointless_const = os.path.join(
                self.get_working_directory(),
                f"{self._scalr_pname}_{self._scalr_xname}_prepointless_const.mtz",
            )
            FileHandler.record_temporary_file(pointless_const)

            aimless_const = self._factory.Aimless()
            aimless_const.set_hklin(pointless_hklin)
            aimless_const.set_hklout(pointless_const)
            aimless_const.const()

            pointless_const = os.path.join(
                self.get_working_directory(),
                "%s_%s_prepointless_const_unmerged.mtz"
                % (self._scalr_pname, self._scalr_xname),
            )
            FileHandler.record_temporary_file(pointless_const)
            pointless_hklin = pointless_const

            pointgroup, reindex_op, ntr, pt = self._pointless_indexer_multisweep(
                pointless_hklin, refiners
            )

            for epoch in self._sweep_handler.get_epochs():
                pointgroups[epoch] = pointgroup
                reindex_ops[epoch] = reindex_op
            # SUMMARY ran pointless multisweep on combined mtz and made a dict
            # of  pointgroups and reindex_ops (all same)
        # END OF if multi-sweep and not input pg

        # START OF if not mulit-sweep or pg given
        else:
            for epoch in self._sweep_handler.get_epochs():
                si = self._sweep_handler.get_sweep_information(epoch)

                hklin = si.get_reflections()

                integrater = si.get_integrater()
                refiner = integrater.get_integrater_refiner()

                if self._scalr_input_pointgroup:
                    logger.debug(
                        "Using input pointgroup: %s", self._scalr_input_pointgroup
                    )
                    pointgroup = self._scalr_input_pointgroup
                    reindex_op = "h,k,l"
                    pt = False

                else:

                    pointless_hklin = self._prepare_pointless_hklin(
                        hklin, si.get_integrater().get_phi_width()
                    )

                    pointgroup, reindex_op, ntr, pt = self._pointless_indexer_jiffy(
                        pointless_hklin, refiner
                    )

                    logger.debug("X1698: %s: %s", pointgroup, reindex_op)

                    if ntr:

                        integrater.integrater_reset_reindex_operator()
                        need_to_return = True

                if pt and not probably_twinned:
                    probably_twinned = True

                logger.debug("Pointgroup: %s (%s)", pointgroup, reindex_op)

                pointgroups[epoch] = pointgroup
                reindex_ops[epoch] = reindex_op
            # SUMMARY - for each sweep, run indexer jiffy and get reindex operators
            # and pointgroups dictionaries (could be different between sweeps)

        # END OF if not mulit-sweep or pg given

        overall_pointgroup = None

        pointgroup_set = {pointgroups[e] for e in pointgroups}

        if len(pointgroup_set) > 1 and not probably_twinned:
            raise RuntimeError(
                "non uniform pointgroups: %s" % str(list(pointgroup_set))
            )

        if len(pointgroup_set) > 1:
            logger.debug(
                "Probably twinned, pointgroups: %s",
                " ".join(p.replace(" ", "") for p in pointgroup_set),
            )
            numbers = (Syminfo.spacegroup_name_to_number(ps) for ps in pointgroup_set)
            overall_pointgroup = Syminfo.spacegroup_number_to_name(min(numbers))
            self._scalr_input_pointgroup = overall_pointgroup

            logger.info("Twinning detected, assume pointgroup %s", overall_pointgroup)

            need_to_return = True

        else:
            overall_pointgroup = pointgroup_set.pop()
        # SUMMARY - Have handled if different pointgroups & chosen an overall_pointgroup
        # which is the lowest symmetry

        # Now go through sweeps and do reindexing
        for epoch in self._sweep_handler.get_epochs():
            si = self._sweep_handler.get_sweep_information(epoch)

            integrater = si.get_integrater()

            integrater.set_integrater_spacegroup_number(
                Syminfo.spacegroup_name_to_number(overall_pointgroup)
            )
            integrater.set_integrater_reindex_operator(
                reindex_ops[epoch], reason="setting point group"
            )
            # This will give us the reflections in the correct point group
            si.set_reflections(integrater.get_integrater_intensities())

        if need_to_return:
            self.set_scaler_done(False)
            self.set_scaler_prepare_done(False)
            return

        # in here now optionally work through the data files which should be
        # indexed with a consistent point group, and transform the orientation
        # matrices by the lattice symmetry operations (if possible) to get a
        # consistent definition of U matrix modulo fixed rotations

        if PhilIndex.params.xia2.settings.unify_setting:
            self.unify_setting()

        if self.get_scaler_reference_reflection_file():
            self._reference = self.get_scaler_reference_reflection_file()
            logger.debug("Using HKLREF %s", self._reference)

        elif PhilIndex.params.xia2.settings.scale.reference_reflection_file:
            self._reference = (
                PhilIndex.params.xia2.settings.scale.reference_reflection_file
            )
            logger.debug("Using HKLREF %s", self._reference)

        params = PhilIndex.params
        use_brehm_diederichs = params.xia2.settings.use_brehm_diederichs
        if len(self._sweep_handler.get_epochs()) > 1 and use_brehm_diederichs:
            self.brehm_diederichs_reindexing()
        # If not Brehm-deidrichs, set reference as first sweep
        elif len(self._sweep_handler.get_epochs()) > 1 and not self._reference:

            first = self._sweep_handler.get_epochs()[0]
            si = self._sweep_handler.get_sweep_information(first)
            self._reference = si.get_reflections()

        # Now reindex to be consistent with first dataset - run pointless on each
        # dataset with reference
        if self._reference:

            md = self._factory.Mtzdump()
            md.set_hklin(self._reference)
            md.dump()

            datasets = md.get_datasets()

            # then get the unit cell, lattice etc.

            reference_lattice = Syminfo.get_lattice(md.get_spacegroup())
            reference_cell = md.get_dataset_info(datasets[0])["cell"]

            # then compute the pointgroup from this...

            # ---------- REINDEX TO CORRECT (REFERENCE) SETTING ----------

            for epoch in self._sweep_handler.get_epochs():

                # if we are working with unified UB matrix then this should not
                # be a problem here (note, *if*; *should*)

                # what about e.g. alternative P1 settings?
                # see JIRA MXSW-904
                if PhilIndex.params.xia2.settings.unify_setting:
                    continue

                pl = self._factory.Pointless()

                si = self._sweep_handler.get_sweep_information(epoch)
                hklin = si.get_reflections()

                pl.set_hklin(
                    self._prepare_pointless_hklin(
                        hklin, si.get_integrater().get_phi_width()
                    )
                )

                hklout = os.path.join(
                    self.get_working_directory(),
                    "%s_rdx2.mtz" % os.path.split(hklin)[-1][:-4],
                )

                # we will want to delete this one exit
                FileHandler.record_temporary_file(hklout)

                # now set the initial reflection set as a reference...

                pl.set_hklref(self._reference)

                # https://github.com/xia2/xia2/issues/115 - should ideally iteratively
                # construct a reference or a tree of correlations to ensure correct
                # reference setting - however if small molecule assume has been
                # multi-sweep-indexed so can ignore "fatal errors" - temporary hack
                pl.decide_pointgroup(
                    ignore_errors=PhilIndex.params.xia2.settings.small_molecule
                )

                logger.debug("Reindexing analysis of %s", pl.get_hklin())

                pointgroup = pl.get_pointgroup()
                reindex_op = pl.get_reindex_operator()

                logger.debug("Operator: %s", reindex_op)

                # apply this...

                integrater = si.get_integrater()

                integrater.set_integrater_reindex_operator(
                    reindex_op, reason="match reference"
                )
                integrater.set_integrater_spacegroup_number(
                    Syminfo.spacegroup_name_to_number(pointgroup)
                )
                si.set_reflections(integrater.get_integrater_intensities())

                md = self._factory.Mtzdump()
                md.set_hklin(si.get_reflections())
                md.dump()

                datasets = md.get_datasets()

                if len(datasets) > 1:
                    raise RuntimeError(
                        "more than one dataset in %s" % si.get_reflections()
                    )

                # then get the unit cell, lattice etc.

                lattice = Syminfo.get_lattice(md.get_spacegroup())
                cell = md.get_dataset_info(datasets[0])["cell"]

                if lattice != reference_lattice:
                    raise RuntimeError(
                        "lattices differ in %s and %s"
                        % (self._reference, si.get_reflections())
                    )

                logger.debug("Cell: %.2f %.2f %.2f %.2f %.2f %.2f" % cell)
                logger.debug("Ref:  %.2f %.2f %.2f %.2f %.2f %.2f" % reference_cell)

                for j in range(6):
                    if (
                        math.fabs((cell[j] - reference_cell[j]) / reference_cell[j])
                        > 0.1
                    ):
                        raise RuntimeError(
                            "unit cell parameters differ in %s and %s"
                            % (self._reference, si.get_reflections())
                        )

        # ---------- SORT TOGETHER DATA ----------

        self._sort_together_data_ccp4()

        self._scalr_resolution_limits = {}

        # store central resolution limit estimates

        batch_ranges = [
            self._sweep_handler.get_sweep_information(epoch).get_batch_range()
            for epoch in self._sweep_handler.get_epochs()
        ]

        self._resolution_limit_estimates = ersatz_resolution(
            self._prepared_reflections, batch_ranges
        )
Пример #5
0
    def _sort_together_data_ccp4(self):
        '''Sort together in the right order (rebatching as we go) the sweeps
    we want to scale together.'''

        max_batches = 0

        for epoch in self._sweep_handler.get_epochs():

            si = self._sweep_handler.get_sweep_information(epoch)
            pname, xname, dname = si.get_project_info()
            sname = si.get_sweep_name()
            hklin = si.get_reflections()

            # limit the reflections - e.g. if we are re-running the scaling step
            # on just a subset of the integrated data

            hklin = si.get_reflections()
            limit_batch_range = None
            for sweep in PhilIndex.params.xia2.settings.sweep:
                if sweep.id == sname and sweep.range is not None:
                    limit_batch_range = sweep.range
                    break

            if limit_batch_range is not None:
                Debug.write('Limiting batch range for %s: %s' %
                            (sname, limit_batch_range))
                start, end = limit_batch_range
                hklout = os.path.splitext(hklin)[0] + '_tmp.mtz'
                FileHandler.record_temporary_file(hklout)
                rb = self._factory.Pointless()
                rb.set_hklin(hklin)
                rb.set_hklout(hklout)
                rb.limit_batches(start, end)
                si.set_reflections(hklout)
                si.set_batches(limit_batch_range)

            # keep a count of the maximum number of batches in a block -
            # this will be used to make rebatch work below.

            hklin = si.get_reflections()

            batches = MtzUtils.batches_from_mtz(hklin)
            if 1 + max(batches) - min(batches) > max_batches:
                max_batches = max(batches) - min(batches) + 1

        Debug.write('Biggest sweep has %d batches' % max_batches)
        max_batches = nifty_power_of_ten(max_batches)

        # then rebatch the files, to make sure that the batch numbers are
        # in the same order as the epochs of data collection.

        counter = 0

        for epoch in self._sweep_handler.get_epochs():

            si = self._sweep_handler.get_sweep_information(epoch)

            hklin = si.get_reflections()

            pname, xname, dname = si.get_project_info()
            sname = si.get_sweep_name()

            hklout = os.path.join(self.get_working_directory(),
                                  '%s_%s_%s_%s_integrated.mtz' % \
                                  (pname, xname, dname, sname))

            first_batch = min(si.get_batches())
            si.set_batch_offset(counter * max_batches - first_batch + 1)

            from xia2.Modules.Scaler.rebatch import rebatch
            new_batches = rebatch(hklin,
                                  hklout,
                                  first_batch=counter * max_batches + 1,
                                  pname=pname,
                                  xname=xname,
                                  dname=dname)

            # update the "input information"

            si.set_reflections(hklout)
            si.set_batches(new_batches)

            # update the counter & recycle

            counter += 1

        s = self._factory.Sortmtz()

        hklout = os.path.join(self.get_working_directory(),
                              '%s_%s_sorted.mtz' % \
                              (self._scalr_pname, self._scalr_xname))

        s.set_hklout(hklout)

        for epoch in self._sweep_handler.get_epochs():
            s.add_hklin(
                self._sweep_handler.get_sweep_information(
                    epoch).get_reflections())

        s.sort()

        # verify that the measurements are in the correct setting
        # choice for the spacegroup

        hklin = hklout
        hklout = hklin.replace('sorted.mtz', 'temp.mtz')

        if not self.get_scaler_reference_reflection_file():

            if PhilIndex.params.xia2.settings.symmetry.program == 'dials':
                p = self._factory.dials_symmetry()
            else:
                p = self._factory.Pointless()

            FileHandler.record_log_file('%s %s pointless' % \
                                        (self._scalr_pname,
                                         self._scalr_xname),
                                        p.get_log_file())

            if len(self._sweep_handler.get_epochs()) > 1:
                p.set_hklin(hklin)
            else:
                # permit the use of pointless preparation...
                epoch = self._sweep_handler.get_epochs()[0]
                p.set_hklin(
                    self._prepare_pointless_hklin(
                        hklin,
                        self._sweep_handler.get_sweep_information(
                            epoch).get_integrater().get_phi_width()))

            if self._scalr_input_spacegroup:
                Debug.write('Assigning user input spacegroup: %s' % \
                            self._scalr_input_spacegroup)

                p.decide_spacegroup()
                spacegroup = p.get_spacegroup()
                reindex_operator = p.get_spacegroup_reindex_operator()

                Debug.write('Pointless thought %s (reindex as %s)' % \
                            (spacegroup, reindex_operator))

                spacegroup = self._scalr_input_spacegroup
                reindex_operator = 'h,k,l'
                self._spacegroup_reindex_operator = reindex_operator

            else:
                p.decide_spacegroup()
                spacegroup = p.get_spacegroup()
                reindex_operator = p.get_spacegroup_reindex_operator()
                self._spacegroup_reindex_operator = clean_reindex_operator(
                    reindex_operator)
                Debug.write('Pointless thought %s (reindex as %s)' % \
                            (spacegroup, reindex_operator))

            if self._scalr_input_spacegroup:
                self._scalr_likely_spacegroups = [self._scalr_input_spacegroup]
            else:
                self._scalr_likely_spacegroups = p.get_likely_spacegroups()

            Chatter.write('Likely spacegroups:')
            for spag in self._scalr_likely_spacegroups:
                Chatter.write('%s' % spag)

            Chatter.write(
                'Reindexing to first spacegroup setting: %s (%s)' % \
                (spacegroup, clean_reindex_operator(reindex_operator)))

        else:
            spacegroup = MtzUtils.space_group_name_from_mtz(
                self.get_scaler_reference_reflection_file())
            reindex_operator = 'h,k,l'

            self._scalr_likely_spacegroups = [spacegroup]

            Debug.write('Assigning spacegroup %s from reference' % \
                        spacegroup)

        # then run reindex to set the correct spacegroup

        ri = self._factory.Reindex()
        ri.set_hklin(hklin)
        ri.set_hklout(hklout)
        ri.set_spacegroup(spacegroup)
        ri.set_operator(reindex_operator)
        ri.reindex()

        FileHandler.record_temporary_file(hklout)

        # then resort the reflections (one last time!)

        s = self._factory.Sortmtz()

        temp = hklin
        hklin = hklout
        hklout = temp

        s.add_hklin(hklin)
        s.set_hklout(hklout)

        s.sort()

        # done preparing!

        self._prepared_reflections = s.get_hklout()
Пример #6
0
    def _sort_together_data_xds(self):

        if len(self._sweep_information) == 1:
            return self._sort_together_data_xds_one_sweep()

        max_batches = 0

        for epoch in self._sweep_information.keys():

            hklin = self._sweep_information[epoch]['scaled_reflections']

            if self._sweep_information[epoch]['batches'] == [0, 0]:

                Chatter.write('Getting batches from %s' % hklin)
                batches = MtzUtils.batches_from_mtz(hklin)
                self._sweep_information[epoch]['batches'] = [
                    min(batches), max(batches)
                ]
                Chatter.write('=> %d to %d' % (min(batches), max(batches)))

            batches = self._sweep_information[epoch]['batches']
            if 1 + max(batches) - min(batches) > max_batches:
                max_batches = max(batches) - min(batches) + 1

        Debug.write('Biggest sweep has %d batches' % max_batches)
        max_batches = nifty_power_of_ten(max_batches)

        epochs = sorted(self._sweep_information.keys())

        counter = 0

        for epoch in epochs:

            hklin = self._sweep_information[epoch]['scaled_reflections']

            pname = self._sweep_information[epoch]['pname']
            xname = self._sweep_information[epoch]['xname']
            dname = self._sweep_information[epoch]['dname']

            sname = self._sweep_information[epoch]['sname']

            hklout = os.path.join(self.get_working_directory(),
                                  '%s_%s_%s_%d.mtz' % \
                                  (pname, xname, dname, counter))

            # we will want to delete this one exit
            FileHandler.record_temporary_file(hklout)

            # record this for future reference - will be needed in the
            # radiation damage analysis...

            # hack - reset this as it gets in a muddle...
            intgr = self._sweep_information[epoch]['integrater']
            self._sweep_information[epoch][
                'batches'] = intgr.get_integrater_batches()

            first_batch = min(self._sweep_information[epoch]['batches'])
            offset = counter * max_batches - first_batch + 1
            self._sweep_information[epoch]['batch_offset'] = offset

            from xia2.Modules.Scaler.rebatch import rebatch
            new_batches = rebatch(hklin,
                                  hklout,
                                  add_batch=offset,
                                  pname=pname,
                                  xname=xname,
                                  dname=dname)

            # update the "input information"

            self._sweep_information[epoch]['hklin'] = hklout
            self._sweep_information[epoch]['batches'] = new_batches

            # update the counter & recycle

            counter += 1

        s = self._factory.Sortmtz()

        hklout = os.path.join(self.get_working_directory(),
                              '%s_%s_sorted.mtz' % \
                              (self._scalr_pname, self._scalr_xname))

        s.set_hklout(hklout)

        for epoch in epochs:
            s.add_hklin(self._sweep_information[epoch]['hklin'])

        s.sort(vrset=-99999999.0)

        self._prepared_reflections = hklout

        if self.get_scaler_reference_reflection_file():
            spacegroups = [
                MtzUtils.space_group_name_from_mtz(
                    self.get_scaler_reference_reflection_file())
            ]
            reindex_operator = 'h,k,l'

        else:
            pointless = self._factory.Pointless()
            pointless.set_hklin(hklout)
            pointless.decide_spacegroup()

            FileHandler.record_log_file('%s %s pointless' % \
                                        (self._scalr_pname,
                                         self._scalr_xname),
                                        pointless.get_log_file())

            spacegroups = pointless.get_likely_spacegroups()
            reindex_operator = pointless.get_spacegroup_reindex_operator()

            if self._scalr_input_spacegroup:
                Debug.write('Assigning user input spacegroup: %s' % \
                            self._scalr_input_spacegroup)
                spacegroups = [self._scalr_input_spacegroup]
                reindex_operator = 'h,k,l'

        self._scalr_likely_spacegroups = spacegroups
        spacegroup = self._scalr_likely_spacegroups[0]

        self._scalr_reindex_operator = reindex_operator

        Chatter.write('Likely spacegroups:')
        for spag in self._scalr_likely_spacegroups:
            Chatter.write('%s' % spag)

        Chatter.write(
            'Reindexing to first spacegroup setting: %s (%s)' % \
            (spacegroup, clean_reindex_operator(reindex_operator)))

        hklin = self._prepared_reflections
        hklout = os.path.join(self.get_working_directory(),
                              '%s_%s_reindex.mtz' % \
                              (self._scalr_pname, self._scalr_xname))

        FileHandler.record_temporary_file(hklout)

        ri = self._factory.Reindex()
        ri.set_hklin(hklin)
        ri.set_hklout(hklout)
        ri.set_spacegroup(spacegroup)
        ri.set_operator(reindex_operator)
        ri.reindex()

        hklin = hklout
        hklout = os.path.join(self.get_working_directory(),
                              '%s_%s_sorted.mtz' % \
                              (self._scalr_pname, self._scalr_xname))

        s = self._factory.Sortmtz()
        s.set_hklin(hklin)
        s.set_hklout(hklout)

        s.sort(vrset=-99999999.0)

        self._prepared_reflections = hklout

        Debug.write(
            'Updating unit cell to %.2f %.2f %.2f %.2f %.2f %.2f' % \
            tuple(ri.get_cell()))
        self._scalr_cell = tuple(ri.get_cell())

        return
Пример #7
0
  def _scale_prepare(self):
    '''Perform all of the preparation required to deliver the scaled
    data. This should sort together the reflection files, ensure that
    they are correctly indexed (via pointless) and generally tidy
    things up.'''

    # acknowledge all of the programs we are about to use...

    Citations.cite('pointless')
    Citations.cite('aimless')
    Citations.cite('ccp4')

    # ---------- GATHER ----------

    self._sweep_handler = SweepInformationHandler(self._scalr_integraters)

    Journal.block(
        'gathering', self.get_scaler_xcrystal().get_name(), 'CCP4',
        {'working directory':self.get_working_directory()})

    for epoch in self._sweep_handler.get_epochs():
      si = self._sweep_handler.get_sweep_information(epoch)
      pname, xname, dname = si.get_project_info()
      sname = si.get_sweep_name()

      exclude_sweep = False

      for sweep in PhilIndex.params.xia2.settings.sweep:
        if sweep.id == sname and sweep.exclude:
          exclude_sweep = True
          break

      if exclude_sweep:
        self._sweep_handler.remove_epoch(epoch)
        Debug.write('Excluding sweep %s' % sname)
      else:
        Journal.entry({'adding data from':'%s/%s/%s' % \
                       (xname, dname, sname)})

    # gather data for all images which belonged to the parent
    # crystal - allowing for the fact that things could go wrong
    # e.g. epoch information not available, exposure times not in
    # headers etc...

    for e in self._sweep_handler.get_epochs():
      si = self._sweep_handler.get_sweep_information(e)
      assert is_mtz_file(si.get_reflections())

    p, x = self._sweep_handler.get_project_info()
    self._scalr_pname = p
    self._scalr_xname = x

    # verify that the lattices are consistent, calling eliminate if
    # they are not N.B. there could be corner cases here

    need_to_return = False

    multi_sweep_indexing = \
      PhilIndex.params.xia2.settings.multi_sweep_indexing == True

    if len(self._sweep_handler.get_epochs()) > 1:

      # if we have multi-sweep-indexing going on then logic says all should
      # share common lattice & UB definition => this is not used here?
      if multi_sweep_indexing and not self._scalr_input_pointgroup:
        pointless_hklins = []

        max_batches = 0
        for epoch in self._sweep_handler.get_epochs():
          si = self._sweep_handler.get_sweep_information(epoch)
          hklin = si.get_reflections()

          batches = MtzUtils.batches_from_mtz(hklin)
          if 1 + max(batches) - min(batches) > max_batches:
            max_batches = max(batches) - min(batches) + 1

        from xia2.lib.bits import nifty_power_of_ten
        Debug.write('Biggest sweep has %d batches' % max_batches)
        max_batches = nifty_power_of_ten(max_batches)

        counter = 0

        refiners = []

        for epoch in self._sweep_handler.get_epochs():
          si = self._sweep_handler.get_sweep_information(epoch)
          hklin = si.get_reflections()
          integrater = si.get_integrater()
          refiner = integrater.get_integrater_refiner()
          refiners.append(refiner)

          hklin = self._prepare_pointless_hklin(
            hklin, si.get_integrater().get_phi_width())

          hklout = os.path.join(self.get_working_directory(),
                                '%s_%s_%s_%s_prepointless.mtz' % \
                                (pname, xname, dname, si.get_sweep_name()))

          # we will want to delete this one exit
          FileHandler.record_temporary_file(hklout)

          first_batch = min(si.get_batches())
          si.set_batch_offset(counter * max_batches - first_batch + 1)

          from xia2.Modules.Scaler.rebatch import rebatch
          new_batches = rebatch(
            hklin, hklout, first_batch=counter * max_batches + 1,
            pname=pname, xname=xname, dname=dname)

          pointless_hklins.append(hklout)

          # update the counter & recycle
          counter += 1

        s = self._factory.Sortmtz()

        pointless_hklin = os.path.join(self.get_working_directory(),
                              '%s_%s_prepointless_sorted.mtz' % \
                              (self._scalr_pname, self._scalr_xname))

        s.set_hklout(pointless_hklin)

        for hklin in pointless_hklins:
          s.add_hklin(hklin)

        s.sort()

        # FIXME xia2-51 in here look at running constant scaling on the
        # pointless hklin to put the runs on the same scale. Ref=[A]

        pointless_const = os.path.join(self.get_working_directory(),
                              '%s_%s_prepointless_const.mtz' % \
                              (self._scalr_pname, self._scalr_xname))
        FileHandler.record_temporary_file(pointless_const)

        aimless_const = self._factory.Aimless()
        aimless_const.set_hklin(pointless_hklin)
        aimless_const.set_hklout(pointless_const)
        aimless_const.const()

        pointless_const = os.path.join(self.get_working_directory(),
                              '%s_%s_prepointless_const_unmerged.mtz' % \
                              (self._scalr_pname, self._scalr_xname))
        FileHandler.record_temporary_file(pointless_const)
        pointless_hklin = pointless_const

        # FIXME xia2-51 in here need to pass all refiners to ensure that the
        # information is passed back to all of them not just the last one...
        Debug.write('Running multisweep pointless for %d sweeps' %
                    len(refiners))
        pointgroup, reindex_op, ntr, pt = \
                    self._pointless_indexer_multisweep(pointless_hklin,
                                                       refiners)

        Debug.write('X1698: %s: %s' % (pointgroup, reindex_op))

        lattices = [Syminfo.get_lattice(pointgroup)]

        for epoch in self._sweep_handler.get_epochs():
          si = self._sweep_handler.get_sweep_information(epoch)
          intgr = si.get_integrater()
          hklin = si.get_reflections()
          refiner = intgr.get_integrater_refiner()

          if ntr:
            intgr.integrater_reset_reindex_operator()
            need_to_return = True

      else:
        lattices = []

        for epoch in self._sweep_handler.get_epochs():

          si = self._sweep_handler.get_sweep_information(epoch)
          intgr = si.get_integrater()
          hklin = si.get_reflections()
          refiner = intgr.get_integrater_refiner()

          if self._scalr_input_pointgroup:
            pointgroup = self._scalr_input_pointgroup
            reindex_op = 'h,k,l'
            ntr = False

          else:
            pointless_hklin = self._prepare_pointless_hklin(
              hklin, si.get_integrater().get_phi_width())

            pointgroup, reindex_op, ntr, pt = \
                        self._pointless_indexer_jiffy(
                pointless_hklin, refiner)

            Debug.write('X1698: %s: %s' % (pointgroup, reindex_op))

          lattice = Syminfo.get_lattice(pointgroup)

          if not lattice in lattices:
            lattices.append(lattice)

          if ntr:

            intgr.integrater_reset_reindex_operator()
            need_to_return = True

      if len(lattices) > 1:

        # why not using pointless indexer jiffy??!

        correct_lattice = sort_lattices(lattices)[0]

        Chatter.write('Correct lattice asserted to be %s' % \
                      correct_lattice)

        # transfer this information back to the indexers
        for epoch in self._sweep_handler.get_epochs():

          si = self._sweep_handler.get_sweep_information(epoch)
          refiner = si.get_integrater().get_integrater_refiner()
          sname = si.get_sweep_name()

          state = refiner.set_refiner_asserted_lattice(
              correct_lattice)

          if state == refiner.LATTICE_CORRECT:
            Chatter.write('Lattice %s ok for sweep %s' % \
                          (correct_lattice, sname))
          elif state == refiner.LATTICE_IMPOSSIBLE:
            raise RuntimeError('Lattice %s impossible for %s' \
                  % (correct_lattice, sname))
          elif state == refiner.LATTICE_POSSIBLE:
            Chatter.write('Lattice %s assigned for sweep %s' % \
                          (correct_lattice, sname))
            need_to_return = True

    # if one or more of them was not in the lowest lattice,
    # need to return here to allow reprocessing

    if need_to_return:
      self.set_scaler_done(False)
      self.set_scaler_prepare_done(False)
      return

    # ---------- REINDEX ALL DATA TO CORRECT POINTGROUP ----------

    # all should share the same pointgroup, unless twinned... in which
    # case force them to be...

    pointgroups = {}
    reindex_ops = {}
    probably_twinned = False

    need_to_return = False

    multi_sweep_indexing = \
      PhilIndex.params.xia2.settings.multi_sweep_indexing == True

    if multi_sweep_indexing and not self._scalr_input_pointgroup:
      pointless_hklins = []

      max_batches = 0
      for epoch in self._sweep_handler.get_epochs():
        si = self._sweep_handler.get_sweep_information(epoch)
        hklin = si.get_reflections()

        batches = MtzUtils.batches_from_mtz(hklin)
        if 1 + max(batches) - min(batches) > max_batches:
          max_batches = max(batches) - min(batches) + 1

      from xia2.lib.bits import nifty_power_of_ten
      Debug.write('Biggest sweep has %d batches' % max_batches)
      max_batches = nifty_power_of_ten(max_batches)

      counter = 0

      refiners = []

      for epoch in self._sweep_handler.get_epochs():
        si = self._sweep_handler.get_sweep_information(epoch)
        hklin = si.get_reflections()
        integrater = si.get_integrater()
        refiner = integrater.get_integrater_refiner()
        refiners.append(refiner)

        hklin = self._prepare_pointless_hklin(
            hklin, si.get_integrater().get_phi_width())

        hklout = os.path.join(self.get_working_directory(),
                              '%s_%s_%s_%s_prepointless.mtz' % \
                              (pname, xname, dname, si.get_sweep_name()))

        # we will want to delete this one exit
        FileHandler.record_temporary_file(hklout)

        first_batch = min(si.get_batches())
        si.set_batch_offset(counter * max_batches - first_batch + 1)

        from xia2.Modules.Scaler.rebatch import rebatch
        new_batches = rebatch(
          hklin, hklout, first_batch=counter * max_batches + 1,
          pname=pname, xname=xname, dname=dname)

        pointless_hklins.append(hklout)

        # update the counter & recycle
        counter += 1

      # FIXME related to xia2-51 - this looks very very similar to the logic
      # in [A] above - is this duplicated logic?
      s = self._factory.Sortmtz()

      pointless_hklin = os.path.join(self.get_working_directory(),
                            '%s_%s_prepointless_sorted.mtz' % \
                            (self._scalr_pname, self._scalr_xname))

      s.set_hklout(pointless_hklin)

      for hklin in pointless_hklins:
        s.add_hklin(hklin)

      s.sort()

      pointless_const = os.path.join(self.get_working_directory(),
                            '%s_%s_prepointless_const.mtz' % \
                            (self._scalr_pname, self._scalr_xname))
      FileHandler.record_temporary_file(pointless_const)

      aimless_const = self._factory.Aimless()
      aimless_const.set_hklin(pointless_hklin)
      aimless_const.set_hklout(pointless_const)
      aimless_const.const()

      pointless_const = os.path.join(self.get_working_directory(),
                            '%s_%s_prepointless_const_unmerged.mtz' % \
                            (self._scalr_pname, self._scalr_xname))
      FileHandler.record_temporary_file(pointless_const)
      pointless_hklin = pointless_const

      pointgroup, reindex_op, ntr, pt = \
                  self._pointless_indexer_multisweep(
          pointless_hklin, refiners)

      for epoch in self._sweep_handler.get_epochs():
        pointgroups[epoch] = pointgroup
        reindex_ops[epoch] = reindex_op

    else:
      for epoch in self._sweep_handler.get_epochs():
        si = self._sweep_handler.get_sweep_information(epoch)

        hklin = si.get_reflections()

        integrater = si.get_integrater()
        refiner = integrater.get_integrater_refiner()

        if self._scalr_input_pointgroup:
          Debug.write('Using input pointgroup: %s' % \
                      self._scalr_input_pointgroup)
          pointgroup = self._scalr_input_pointgroup
          reindex_op = 'h,k,l'
          pt = False

        else:

          pointless_hklin = self._prepare_pointless_hklin(
              hklin, si.get_integrater().get_phi_width())

          pointgroup, reindex_op, ntr, pt = \
                      self._pointless_indexer_jiffy(
              pointless_hklin, refiner)

          Debug.write('X1698: %s: %s' % (pointgroup, reindex_op))

          if ntr:

            integrater.integrater_reset_reindex_operator()
            need_to_return = True

        if pt and not probably_twinned:
          probably_twinned = True

        Debug.write('Pointgroup: %s (%s)' % (pointgroup, reindex_op))

        pointgroups[epoch] = pointgroup
        reindex_ops[epoch] = reindex_op

    overall_pointgroup = None

    pointgroup_set = {pointgroups[e] for e in pointgroups}

    if len(pointgroup_set) > 1 and \
       not probably_twinned:
      raise RuntimeError('non uniform pointgroups')

    if len(pointgroup_set) > 1:
      Debug.write('Probably twinned, pointgroups: %s' % \
                  ' '.join([p.replace(' ', '') for p in \
                            list(pointgroup_set)]))
      numbers = [Syminfo.spacegroup_name_to_number(s) for s in \
                 pointgroup_set]
      overall_pointgroup = Syminfo.spacegroup_number_to_name(min(numbers))
      self._scalr_input_pointgroup = overall_pointgroup

      Chatter.write('Twinning detected, assume pointgroup %s' % \
                    overall_pointgroup)

      need_to_return = True

    else:
      overall_pointgroup = pointgroup_set.pop()

    for epoch in self._sweep_handler.get_epochs():
      si = self._sweep_handler.get_sweep_information(epoch)

      integrater = si.get_integrater()

      integrater.set_integrater_spacegroup_number(
          Syminfo.spacegroup_name_to_number(overall_pointgroup))
      integrater.set_integrater_reindex_operator(
          reindex_ops[epoch], reason='setting point group')
      # This will give us the reflections in the correct point group
      si.set_reflections(integrater.get_integrater_intensities())

    if need_to_return:
      self.set_scaler_done(False)
      self.set_scaler_prepare_done(False)
      return

    # in here now optionally work through the data files which should be
    # indexed with a consistent point group, and transform the orientation
    # matrices by the lattice symmetry operations (if possible) to get a
    # consistent definition of U matrix modulo fixed rotations

    if PhilIndex.params.xia2.settings.unify_setting:

      from scitbx.matrix import sqr
      reference_U = None
      i3 = sqr((1, 0, 0, 0, 1, 0, 0, 0, 1))

      for epoch in self._sweep_handler.get_epochs():
        si = self._sweep_handler.get_sweep_information(epoch)
        intgr = si.get_integrater()
        fixed = sqr(intgr.get_goniometer().get_fixed_rotation())
        u, b, s = get_umat_bmat_lattice_symmetry_from_mtz(si.get_reflections())
        U = fixed.inverse() * sqr(u).transpose()
        B = sqr(b)

        if reference_U is None:
          reference_U = U
          continue

        results = []
        for op in s.all_ops():
          R = B * sqr(op.r().as_double()).transpose() * B.inverse()
          nearly_i3 = (U * R).inverse() * reference_U
          score = sum([abs(_n - _i) for (_n, _i) in zip(nearly_i3, i3)])
          results.append((score, op.r().as_hkl(), op))

        results.sort()
        best = results[0]
        Debug.write('Best reindex: %s %.3f' % (best[1], best[0]))
        intgr.set_integrater_reindex_operator(best[2].r().inverse().as_hkl(),
                                              reason='unifying [U] setting')
        si.set_reflections(intgr.get_integrater_intensities())

        # recalculate to verify
        u, b, s = get_umat_bmat_lattice_symmetry_from_mtz(si.get_reflections())
        U = fixed.inverse() * sqr(u).transpose()
        Debug.write('New reindex: %s' % (U.inverse() * reference_U))

        # FIXME I should probably raise an exception at this stage if this
        # is not about I3...

    if self.get_scaler_reference_reflection_file():
      self._reference = self.get_scaler_reference_reflection_file()
      Debug.write('Using HKLREF %s' % self._reference)

    elif PhilIndex.params.xia2.settings.scale.reference_reflection_file:
      self._reference = PhilIndex.params.xia2.settings.scale.reference_reflection_file
      Debug.write('Using HKLREF %s' % self._reference)

    params = PhilIndex.params
    use_brehm_diederichs = params.xia2.settings.use_brehm_diederichs
    if len(self._sweep_handler.get_epochs()) > 1 and use_brehm_diederichs:

      brehm_diederichs_files_in = []
      for epoch in self._sweep_handler.get_epochs():

        si = self._sweep_handler.get_sweep_information(epoch)
        hklin = si.get_reflections()
        brehm_diederichs_files_in.append(hklin)

      # now run cctbx.brehm_diederichs to figure out the indexing hand for
      # each sweep
      from xia2.Wrappers.Cctbx.BrehmDiederichs import BrehmDiederichs
      from xia2.lib.bits import auto_logfiler
      brehm_diederichs = BrehmDiederichs()
      brehm_diederichs.set_working_directory(self.get_working_directory())
      auto_logfiler(brehm_diederichs)
      brehm_diederichs.set_input_filenames(brehm_diederichs_files_in)
      # 1 or 3? 1 seems to work better?
      brehm_diederichs.set_asymmetric(1)
      brehm_diederichs.run()
      reindexing_dict = brehm_diederichs.get_reindexing_dict()

      for epoch in self._sweep_handler.get_epochs():

        si = self._sweep_handler.get_sweep_information(epoch)
        intgr = si.get_integrater()
        hklin = si.get_reflections()

        reindex_op = reindexing_dict.get(os.path.abspath(hklin))
        assert reindex_op is not None

        if 1 or reindex_op != 'h,k,l':
          # apply the reindexing operator
          intgr.set_integrater_reindex_operator(
            reindex_op, reason='match reference')
          si.set_reflections(intgr.get_integrater_intensities())

    elif len(self._sweep_handler.get_epochs()) > 1 and \
           not self._reference:

      first = self._sweep_handler.get_epochs()[0]
      si = self._sweep_handler.get_sweep_information(first)
      self._reference = si.get_reflections()

    if self._reference:

      md = self._factory.Mtzdump()
      md.set_hklin(self._reference)
      md.dump()

      if md.get_batches() and False:
        raise RuntimeError('reference reflection file %s unmerged' % \
              self._reference)

      datasets = md.get_datasets()

      if len(datasets) > 1 and False:
        raise RuntimeError('more than one dataset in %s' % \
              self._reference)

      # then get the unit cell, lattice etc.

      reference_lattice = Syminfo.get_lattice(md.get_spacegroup())
      reference_cell = md.get_dataset_info(datasets[0])['cell']

      # then compute the pointgroup from this...

      # ---------- REINDEX TO CORRECT (REFERENCE) SETTING ----------

      for epoch in self._sweep_handler.get_epochs():

        # if we are working with unified UB matrix then this should not
        # be a problem here (note, *if*; *should*)

        # what about e.g. alternative P1 settings?
        # see JIRA MXSW-904
        if PhilIndex.params.xia2.settings.unify_setting:
          continue

        pl = self._factory.Pointless()

        si = self._sweep_handler.get_sweep_information(epoch)
        hklin = si.get_reflections()

        pl.set_hklin(self._prepare_pointless_hklin(
            hklin, si.get_integrater().get_phi_width()))

        hklout = os.path.join(
            self.get_working_directory(),
            '%s_rdx2.mtz' % os.path.split(hklin)[-1][:-4])

        # we will want to delete this one exit
        FileHandler.record_temporary_file(hklout)

        # now set the initial reflection set as a reference...

        pl.set_hklref(self._reference)

        # https://github.com/xia2/xia2/issues/115 - should ideally iteratively
        # construct a reference or a tree of correlations to ensure correct
        # reference setting - however if small molecule assume has been
        # multi-sweep-indexed so can ignore "fatal errors" - temporary hack
        pl.decide_pointgroup(
          ignore_errors=PhilIndex.params.xia2.settings.small_molecule)

        Debug.write('Reindexing analysis of %s' % pl.get_hklin())

        pointgroup = pl.get_pointgroup()
        reindex_op = pl.get_reindex_operator()

        Debug.write('Operator: %s' % reindex_op)

        # apply this...

        integrater = si.get_integrater()

        integrater.set_integrater_reindex_operator(reindex_op,
                                                   reason='match reference')
        integrater.set_integrater_spacegroup_number(
            Syminfo.spacegroup_name_to_number(pointgroup))
        si.set_reflections(integrater.get_integrater_intensities())

        md = self._factory.Mtzdump()
        md.set_hklin(si.get_reflections())
        md.dump()

        datasets = md.get_datasets()

        if len(datasets) > 1:
          raise RuntimeError('more than one dataset in %s' % \
                si.get_reflections())

        # then get the unit cell, lattice etc.

        lattice = Syminfo.get_lattice(md.get_spacegroup())
        cell = md.get_dataset_info(datasets[0])['cell']

        if lattice != reference_lattice:
          raise RuntimeError('lattices differ in %s and %s' % \
                (self._reference, si.get_reflections()))

        Debug.write('Cell: %.2f %.2f %.2f %.2f %.2f %.2f' % cell)
        Debug.write('Ref:  %.2f %.2f %.2f %.2f %.2f %.2f' % reference_cell)

        for j in range(6):
          if math.fabs((cell[j] - reference_cell[j]) /
                       reference_cell[j]) > 0.1:
            raise RuntimeError( \
                  'unit cell parameters differ in %s and %s' % \
                  (self._reference, si.get_reflections()))

    # ---------- SORT TOGETHER DATA ----------

    self._sort_together_data_ccp4()

    self._scalr_resolution_limits = {}

    # store central resolution limit estimates

    batch_ranges = [
        self._sweep_handler.get_sweep_information(epoch).get_batch_range()
        for epoch in self._sweep_handler.get_epochs()
    ]

    self._resolution_limit_estimates = ersatz_resolution(
        self._prepared_reflections, batch_ranges)