This is an update of Nile's 2019 datafix script that uploads Neuropixels ccf infomation from ccf_regions.csv files to the LIMS2 ecephys_channels table

202203110909_pbs1874_datafix.rb

class Pbs1874Datafix < AIBS::Datafix

  # https://github.com/AllenInstitute/AllenSDK/issues/992
  # http://stash.corp.alleninstitute.org/projects/TECH/repos/lims/browse/db/datafixes/20190911134728_pbs1874_datafix.rb?at=ac873b70756defe209124330f8723f1e99de03c3
  #create_ldds_row # this mandatory statement belongs at the beginning of the script since part of the purpose of this class is to determine if your datafix script has already been run. (see http://confluence.corp.alleninstitute.org/display/IT/datafix).

  structures = {}
  acro_id = {}
  isocortical = []
  layer = Set.new

  layer_re = /.*[Ll]ayer.*\d+.*/

  StructureGraph.find(1).graph_as_list().each do |st|
    stid = st["id"].to_i

    if st["path"].include?(",315,")
      isocortical << stid
      if !layer_re.match(st["name"]).nil?
        layer.add(stid)
      end
    end

    structures[stid] = st
    acro_id[st["acronym"]] = stid

  end

  unlayer_map = {}

  isocortical.each do |stid|
    if layer.include?(stid)
      structure = structures.fetch(stid)
      path = structure["path"]
        .slice(1..structure["path"].length-1)
        .split(",")
        .reverse
        .map {|x| x.to_i}

      path.each do |ancestor_id|
        if !layer.include?(ancestor_id)
          unlayer_map[stid] = ancestor_id
          break
        end
      end

    end
  end

  parent_dir = "/allen/programs/braintv/workgroups/neuralcoding/corbettb/VBN_production"
  dir_glob = File.join(parent_dir, "**/ccf_regions.csv")

  dir_re = /#{parent_dir}\/(?<session_id>\d+)_\d+_\d+_(?<probe_name>probe[a-z|A-Z])_sorted.*/

  Dir.glob(dir_glob).each do |path|

    match = dir_re.match(path)

    if match.nil?
      raise "wat"
    end

    session_id = match.named_captures["session_id"].to_i
    probe_name = match.named_captures["probe_name"]

    probes = EcephysProbe.where(ecephys_session_id: session_id, name: probe_name)
    if probes.length != 1
      puts "unable to find #{probe_name} for session #{session_id}"
      next
    end
    probe = probes.first

    table = CSV.table(path)
    table.each do |row|
      if !row.is_a?(Hash)
        row = row.to_hash
      end

      #local_index = row.fetch(nil).to_i
      local_index = row.fetch(:channels).to_i
      channel = EcephysChannel.where(local_index: local_index, ecephys_probe_id: probe.id).first

      if channel.nil?
        puts "no channel at local index #{local_index} for probe #{probe.id} (#{probe.name}) on session #{probe.ecephys_session_id}"
        next
      end

      structure_acronym = row.fetch(:structure_acronym)
#      if !/.*6$/.match(structure_acronym).nil?
#        structure_acronym = structure_acronym + "a"
#      end

      if structure_acronym != 'none'
        structure_id = acro_id.fetch(structure_acronym)
        structure_id = unlayer_map.fetch(structure_id, structure_id)
        channel.manual_structure_id = structure_id
      else
        channel.manual_structure_id = nil
      end

      ap = row.fetch(:ap).to_f * 1000
      dv = row.fetch(:dv).to_f * 1000
      lr = row.fetch(:ml).to_f * 1000

      depth = row.fetch(:cortical_depth).to_f

      channel.anterior_posterior_ccf_coordinate = ap
      channel.dorsal_ventral_ccf_coordinate = dv
      channel.left_right_ccf_coordinate = lr
      channel.manual_cortical_depth = depth

      channel.save!

    end

  end
  #end # end ActiveRecord::Base.transaction

end