Login
Login

Module:Sandbox/genewiki/testdislua

local p = { }

local navbar = require('Module:Navbar')._navbar

local infobox = require('Module:Infobox3cols').infobox

local infoboxImage = require('Module:InfoboxImage').InfoboxImage

--on a page {{#invoke:Sandbox/genewiki/alllua|getTemplateData|QID=Q14865053}}

--in debug window

--frame = mw.getCurrentFrame()

--frame.args = {QID="Q14865053"} Q18031325

--print(p.getTemplateData(frame))

p.getTemplateData = function(frame)

--make some guesses about whether the provided QID is a good one

--could expand here if we had some kind of error handling framework

--did we get it from the page

local root_qid = mw.text.trim(frame.args['QID'] or "") --try to get it from the args

local mm_qid = ""

--pull all the entity objects that we will need

local entity = ""

local entity_protein = {}

local entity_mouse = ""

local entity_mouse_protein = {}

local checkOrtholog = "" --flag used to see if mouse data avaliable

local entity_genasscn = ""

local mouse_propertyID = "P684"

local protein_propertyID = "P688"

local genasscn_propertyID = "P2293"

--get root gene entity

if root_qid == "" then

entity = mw.wikibase.getEntityObject()

if entity then root_qid = entity.id else root_qid = "" end

else

--assuming we think its good make one call to retrieve and store its wikidata representation

entity = mw.wikibase.getEntity(root_qid)

end

--get the other related entities

if entity then

local claims = ""

--get protein entity object

if entity.claims then

claims = entity.claims[protein_propertyID]

end

if claims then

--go through each index and then make entity_protein indexed

if (claims[1] and claims[1].mainsnak.snaktype == "value" and claims[1].mainsnak.datavalue.type == "wikibase-entityid") then

for k, v in pairs(claims) do

local protein_itemID = "Q" .. claims[#entity_protein + 1].mainsnak.datavalue.value["numeric-id"]

entity_protein[#entity_protein + 1] = mw.wikibase.getEntity(protein_itemID)

end

end --will return nothing if no claims are found

end

--get mouse entity object

if entity.claims then

claims = entity.claims[mouse_propertyID]

end

if claims then

if (claims[1] and claims[1].mainsnak.snaktype == "value" and claims[1].mainsnak.datavalue.type == "wikibase-entityid") then

local mouse_itemID = "Q" .. claims[1].mainsnak.datavalue.value["numeric-id"]

mm_qid = mouse_itemID

entity_mouse = mw.wikibase.getEntity(mouse_itemID)

checkOrtholog = 1

end --will return nothing if no claims are found

else

checkOrtholog = 0

end

--get genetic association object

if entity.claims then

claims = entity.claims[genasscn_propertyID]

end

if claims then

if (claims[1] and claims[1].mainsnak.snaktype == "value" and claims[1].mainsnak.datavalue.type == "wikibase-entityid") then

local genasscn_itemID = "Q" .. claims[1].mainsnak.datavalue.value["numeric-id"]

--mm_qid = genasscn_itemID

entity_genasscn = mw.wikibase.getEntity(genasscn_itemID)

end

end

--get mouse protein entity object

if entity_mouse and entity_mouse.claims then

claims = entity_mouse.claims[protein_propertyID]

end

if claims then

if (claims[1] and claims[1].mainsnak.snaktype == "value" and claims[1].mainsnak.datavalue.type == "wikibase-entityid") then

for k, v in pairs(claims) do

local protein_itemID = "Q" .. claims[#entity_mouse_protein + 1].mainsnak.datavalue.value["numeric-id"]

entity_mouse_protein[#entity_mouse_protein + 1] = mw.wikibase.getEntity(protein_itemID)

end

end --will return nothing if no claims are found

end

end

if entity then --only require the main gene entity

--a list variables of all the data in the info box

local name = p.getLabel(entity)

local entrez_gene = p.getValue(entity, "P351")

local entrez_gene_mm = p.getValue(entity_mouse, "P351", "n/a")

local image = p.getImage(entity, "P18", " ", "250px") --need to set size

local uniprotID_hs = p.getValueProtein(entity_protein, "P352", "n/a")

local uniprotID_mm = p.getValueProtein(entity_mouse_protein, "P352", "n/a")

local pdbIDs = p.getPDB(entity_protein) --makes a list with links to RCSB

local aliases = p.getAliases(entity)

local gene_symbol = p.getValue(entity, "P353")

local hgnc_id = p.getValue(entity, "P354")

local homologene_id = p.getValue(entity, "P593")

local omim_id = p.getValue(entity, "P492")

local mgi_id = p.getValue(entity_mouse, "P671")

local ChEMBL_id = ""

local ec_no = ""

local mol_funct = p.getGO(entity_protein, "P680")

local cell_comp = p.getGO(entity_protein, "P681")

local bio_process = p.getGO(entity_protein, "P682")

local expression_images = p.getImage(entity,"P692","

","250px")

local ensembl = p.getValue(entity, "P594", "n/a")

local ensembl_mm = p.getValue(entity_mouse, "P594", "n/a")

local refseq_mRNA = p.getRefseq_mRNA(entity, "P639", "n/a")

local refseq_mRNA_mm = p.getRefseq_mRNA(entity_mouse, "P639", "n/a")

local refseq_prot = p.getRefseq_protein(entity_protein, "P637", "n/a")

local refseq_prot_mm = p.getRefseq_protein(entity_mouse_protein, "P637", "n/a")

local gstart = p.getChromosomeLoc(entity, "P644", "hg")

local gend = p.getChromosomeLoc(entity, "P645", "hg")

local chr = p.trimChromosome(entity)

local db = p.getAliasFromGenomeAssembly(entity,"hg")

local gstart_mm = p.getChromosomeLoc(entity_mouse, "P644", "mm")

local gend_mm = p.getChromosomeLoc(entity_mouse, "P645", "mm")

local db_mm = p.getAliasFromGenomeAssembly(entity_mouse,"mm")

local chr_mm = p.trimChromosome(entity_mouse)

local disease = p.getValue(entity, "P2293")

local disomim_id = p.getValue(entity_genasscn, "P492")

--define Global Color Scheme

rowBGcolor = '#eee'

titleBGcolor = '#ddd'

sideTitleBGcolor = '#c3fdb8'

p.createTable()

p.renderUpperTitle(name)

--p.renderCaption()

p.renderImage(image)

p.renderAvailableStructures(uniprotID_hs, uniprotID_mm, checkOrtholog, pdbIDs) --PDB info

p.renderIdentifiers(aliases, hgnc_id, gene_symbol, homologene_id, omim_id, mgi_id, ChEMBL_id, ec_no, entrez_gene)

p.renderDiseases(entrez_gene, entrez_gene_mm, ensembl, ensembl_mm, uniprotID_hs, uniprotID_mm, refseq_mRNA, refseq_mRNA_mm, refseq_prot, refseq_prot_mm, db, chr, gstart, gend, db_mm, chr_mm, gstart_mm, gend_mm, disease, disomim_id)

if (mol_funct ~= "" and cell_comp ~= "" and bio_process ~= "") then

p.renderGeneOntology(mol_funct, cell_comp, bio_process, uniprotID_hs)

end

if expression_images ~= "" then

p.renderRNAexpression(expression_images, entrez_gene)

end

p.renderOrthologs(entrez_gene, entrez_gene_mm, ensembl, ensembl_mm, uniprotID_hs, uniprotID_mm, refseq_mRNA, refseq_mRNA_mm, refseq_prot, refseq_prot_mm, db, chr, gstart, gend, db_mm, chr_mm, gstart_mm, gend_mm)

p.renderFooter(root_qid, mm_qid)

return tostring(root)

else return "root entity not set"

end

end

p.createTable = function(subbox)

if subbox == 'sub' then --doesn't work

root

:tag('table')

:css('padding', '0')

:css('border', 'none')

:css('margin', '0')

:css('width', 'auto')

:css('min-width', '100%')

:css('font-size', '100%')

:css('clear', 'none')

:css('float', 'none')

:css('background-color', 'transparent')

else

root = mw.html.create('table')

root

:addClass('infobox')

:css('width', '26.4em')

:newline()

end

end

--Title above image

p.renderUpperTitle = function(name)

local title = name

if not title then return "error: failed to get label"; end

root

:tag('tr')

:tag('th')

:attr('colspan', 4)

:css('text-align', 'center')

:css('font-size', '125%')

:css('font-weight', 'bold')

:wikitext(title)

:newline()

end

--This is a place holder for the image caption, which is stored in wikicommons comments unsure how to access

p.renderCaption = function(entity)

--caption

end

--gets default image

p.renderImage = function(image)

root

:tag('tr')

:tag('td')

:attr('colspan', 4)

:css('text-align', 'center')

:wikitext(image)

:done()

:newline()

end

p.renderAvailableStructures = function(uniprotID_hs, uniprotID_mm, checkOrtholog, pdbIDs)

local title = 'Available structures'

local pdb_link = "PDB"

local searchTitle = ""

local listTitle = "List of PDB id codes"

local PDBe_base = 'http://www.ebi.ac.uk/pdbe/searchResults.html?display=both&term='

local RCSB_base = 'http://www.rcsb.org/pdb/search/smartSubquery.do?smartSearchSubtype=UpAccessionIdQuery&accessionIdList='

local url_uniprot = " "

if checkOrtholog == 1 and uniprotID_mm ~= 'n/a' then

searchTitle = 'Ortholog search: '

url_uniprot = uniprotID_mm

else

searchTitle = 'Human UniProt search: '

url_uniprot = uniprotID_hs

end

local PDBe_list = " " --create a list with " or " if there is more than one uniprot

--get first uniprot in a list

if url_uniprot:match("([^,]+),") then--first check if there is a list if not just assume one value

PDBe_list = string.gsub(url_uniprot, ",", "%%20or%%20") --add or's inststead of commas

else

PDBe_list = url_uniprot

end

local PDBe = "["..PDBe_base..PDBe_list.." PDBe] "

local RCSB = "["..RCSB_base..url_uniprot.." RCSB] "

if string.match(pdbIDs, '%w+') then --if there aren't any PDB_ID don't display this part of the infobox

--p.formatRow(title)---how to not close the tags is a mystery and I could condense code once I figure out

root

:tag('tr')

:tag('td')

:attr('colspan', 4)

:css('text-align', 'center')

:css('background-color', rowBGcolor)

:newline()

--p.createTable('sub')

:tag('table')

:css('padding', '0')

:css('border', 'none')

:css('margin', '0')

:css('width', '100%')

:css('text-align', 'left')

:newline()

:tag('tr') --create title header

:tag('th')

:attr('colspan', '4')

:css('text-align', 'center')

:css('background-color',titleBGcolor)

:wikitext(title)

:done()

:newline()

--p.rowLabel(pdb_link)

:tag('tr')

:tag('th')

:attr('rowspan', '2')

:css('background-color', sideTitleBGcolor)

:css('width', '43px')

:wikitext(pdb_link)

:done()

:tag('td')

:attr('colspan', '2')

:css('background-color', rowBGcolor)

:wikitext(searchTitle)

:tag('span')

:attr('class', 'plainlinks')

:wikitext(PDBe)

:wikitext(RCSB)

:done()

:done()

:done() --this may not be needed

:newline()

--new row for collapsible list of PDB codes

:tag('tr')

:tag('td')

:tag('table')

:attr('class', 'collapsible collapsed')

:css('padding', '0')

:css('border', 'none')

:css('margin', '0')

:css('width', '100%')

:css('text-align', 'left')

:newline()

:tag('tr')

:css('background-color',titleBGcolor)

:css('text-align', 'center')

:newline()

:tag('th')

:attr('colspan', '2')

:wikitext(listTitle)

:done()

:newline()

:done()

:tag('tr')

:tag('td')

:attr('colspan', '2')

:css('background-color', rowBGcolor)

:newline()

:tag('p')

:tag('span')

:attr('class', 'plainlinks')

:wikitext(pdbIDs)

:done()

:newline()

else

return ""

end

end

p.renderIdentifiers = function(aliases, hgnc_id, gene_symbol, homologene_id, omim_id, mgi_id, ChEMBL_id, ec_no, entrez_gene)

local title = "Identifiers"

local label_aliases = "[https://en.wikipedia.org/wiki/Human_Genome_Organisation Aliases]"

local symbol_url

if gene_symbol == "" or gene_symbol == nil then

symbol_url = ""

else

if hgnc_id == "" or hgnc_id == nil then

symbol_url = gene_symbol

else

symbol_url = "[http://www.genenames.org/cgi-bin/gene_symbol_report?hgnc_id="..hgnc_id.." "..gene_symbol.."], "

end

end

aliases = string.gsub(aliases, gene_symbol, "") --get rid of gene name if in aliases list

aliases = string.gsub(aliases, ", ,", ",") --remove comma from middle

aliases = string.gsub(aliases, ", $", "") --remove comma from end

aliases = string.gsub(aliases, "^,", "") --remove comma from start

local label_ext_id = "External IDs"

local omim = ""

if string.match(omim_id, '%w+') then

omim = "OMIM:".." ".."[https://omim.org/entry/"..omim_id.." "..omim_id.."] "

end

local homolo =""

if string.match(homologene_id, '%w+') then

homolo = "HomoloGene:".." ".."[http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=homologene&dopt=HomoloGene&list_uids="..homologene_id.." "..homologene_id.."] "

end

local genecards = ""

if string.match(entrez_gene, '%w+') then

genecards = "GeneCards:".." ".."[http://www.genecards.org/cgi-bin/carddisp.pl?id_type=entrezgene&id="..entrez_gene.." "..entrez_gene.."] "

end

local mgi = ""

if string.match(mgi_id, '%w+') then

mgi = "MGI:".." ".."[http://www.informatics.jax.org/marker/MGI:"..mgi_id.." "..mgi_id.."] "

end

local ChEMBL = ""

if string.match(ChEMBL_id, '%w+') then

ChEMBL = "ChEMBL:".." ".."[https://www.ebi.ac.uk/chembldb/index.php/target/inspect/CHEMBL"..ChEMBL_id.." "..ChEMBL_id.."] "

end

root

:tag('tr')

:tag('th')

:attr('colspan', '4')

:css('text-align', 'center')

:css('background-color', titleBGcolor)

:wikitext(title)

:newline()

:tag('tr')

:tag('th')

:attr('scope', 'row')

:css('background-color', sideTitleBGcolor)

:tag('span')

:attr('class', 'plainlinks')

:wikitext(label_aliases)

:done()

:newline()

:tag('td')

:attr('colspan','3')

:css('background', rowBGcolor)

:tag('span')

:attr('class', 'plainlinks')

:wikitext(symbol_url)

:done()

:wikitext(aliases)

:done()

:newline()

:tag('tr')

:tag('th')

:attr('scope', 'row')

:css('background-color', sideTitleBGcolor)

:wikitext(label_ext_id)

:done()

:newline()

:tag('td')

:attr('colspan', '3')

:css('background-color', rowBGcolor)

:tag('span')

:attr('class', 'plainlinks')

:wikitext(omim)

:wikitext(mgi)

:wikitext(homolo)

:wikitext(ChEMBL)

:wikitext(genecards)

:done()

:newline()

:done()

end

p.renderDiseases = function(entrez_gene, entrez_gene_mm, ensembl, ensembl_mm, uniprot, uniprot_mm, refseq_mRNA, refseq_mRNA_mm, refseq_prot, refseq_prot_mm, db, chr, gstart, gend, db_mm, chr_mm,gstart_mm, gend_mm, disease, disomim_id)

local title = "Disease"

local entrezTitle = "Entrez"

--check all values to see if exists so don't create nonsense links in infobox must have some alphanumberic character

local entrez_link = ""

if string.match(entrez_gene, '%w+') and entrez_gene ~= "n/a" then

entrez_link = "[http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=gene&cmd=retrieve&dopt=default&list_uids="..entrez_gene.."&rn=1 "..entrez_gene.."]"

end

local entrez_mm_link = entrez_gene_mm

if string.match(entrez_gene_mm, '%w+') and entrez_gene_mm ~= "n/a" then

entrez_mm_link = "[http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=gene&cmd=retrieve&dopt=default&list_uids="..entrez_gene_mm.."&rn=1 "..entrez_gene_mm.."]"

end

local shipType = ""

if disease ~= "" then

shipType = "genetic"

end

--local split_omim = mw.text.split(disomim_id,"%s*,%s*")

local split_omim = mw.text.split(disomim_id,",")

-- INCORRECT: entrez_gene = "[https://omim.org/entry/"..split_omim[1].. "|" .. "OMIM entry" .."] "

entrez_gene = "[https://omim.org/entry/"..split_omim[1].." ".."OMIM entry".."] "

disease = ""..disease..""

--entrez_gene = split_omim[1]

-- result[0] will give "cat", result[1] will give "dog"

root

:tag('tr')

:tag('th')

:attr('colspan', '4')

:css('text-align', 'center')

:css('background-color', titleBGcolor)

:wikitext(title)

:done()

:newline()

:tag('tr')

:tag('th')

:wikitext("Disease Name")

:done()

:newline()

:tag('td')

:wikitext("Relationship Type")

:done()

:newline()

:tag('td')

:wikitext("Data Source")

:done()

:newline()

:done()

---this section would be nice to do in a loop if the tag closures were figured out

:tag('tr')

:tag('th')

:attr('scope', 'row')

:wikitext(disease)

:done()

:newline()

:tag('td')

:tag('span')

:wikitext(shipType)

:done()

:newline()

:tag('td')

:tag('span')

:attr('class', 'plainlinks')

:wikitext(entrez_gene)--disomim_id[0])

:done()

:newline()

end

p.renderGeneOntology = function(mol_funct, cell_comp, bio_process, uniprotID)

local title = "Gene ontology"

local mol_funct_title = "Molecular function"

local cell_comp_title = "Cellular component"

local bio_process_title = "Biological process"

local amigo_link = "[http://amigo.geneontology.org/" .. " Amigo]"

local quickGO_link = "[http://www.ebi.ac.uk/QuickGO/" .. " QuickGO]"

root

:tag('tr')

:tag('td')

:attr('colspan', 4)

:css('text-align', 'center')

:css('background-color', rowBGcolor)

:newline()

--p.createTable('sub')

:tag('table')

:attr('class', 'collapsible collapsed')

:css('padding', '0')

:css('border', 'none')

:css('margin', '0')

:css('width', '100%')

:css('text-align', 'left')

:newline()

:tag('tr') --create title bar

:tag('th')

:attr('colspan', '4')

:css('text-align', 'center')

:css('background-color', titleBGcolor)

:wikitext(title)

:done()

:newline()

:tag('tr')

:tag('th')

:css('background-color', sideTitleBGcolor)

:wikitext(mol_funct_title)

:done()

:tag('td')

:css('background-color', rowBGcolor)

:newline()

:tag('span')

:attr('class', 'plainlinks')

:wikitext(mol_funct)

:done()

:newline()

:done()

:tag('tr')

:tag('th')

:css('background-color', sideTitleBGcolor)

:wikitext(cell_comp_title)

:done()

:tag('td')

:css('background-color', rowBGcolor)

:newline()

:tag('span')

:attr('class', 'plainlinks')

:wikitext(cell_comp)

:done()

:newline()

:done()

:tag('tr')

:tag('th')

:css('background-color', sideTitleBGcolor)

:wikitext(bio_process_title)

:done()

:tag('td')

:css('background-color', rowBGcolor)

:newline()

:tag('span')

:attr('class', 'plainlinks')

:wikitext(bio_process)

:done()

:newline()

:done()

:tag('tr')

:tag('td')

:css('background-color', rowBGcolor)

:css('text-align', 'center')

:attr('colspan', '4')

:wikitext("Sources:")

:wikitext(amigo_link)

:wikitext(" / ")

:wikitext(quickGO_link)

:done()

:done()

end

p.renderRNAexpression = function(expression_images, entrez_gene)

local title = "RNA expression pattern"

local biogps_link = "[http://biogps.org/gene/"..entrez_gene.."/ More reference expression data]"

root

:tag('tr')

:tag('th')

:attr('colspan', '4')

:css('text-align', 'center')

:css('background-color', titleBGcolor)

:wikitext(title)

:done()

:newline()

:tag('tr')

:tag('td')

:attr('colspan', '4')

:css('text-align', 'center')

:css('background-color', rowBGcolor)

:wikitext(expression_images)

:done()

:newline()

:tag('tr')

:tag('td')

:attr('colspan', '4')

:css('text-align', 'center')

:css('background-color', rowBGcolor)

:tag('span')

:attr('class', 'plainlinks')

:wikitext(biogps_link)

:done()

:newline()

end

p.renderOrthologs = function(entrez_gene, entrez_gene_mm, ensembl, ensembl_mm, uniprot, uniprot_mm, refseq_mRNA, refseq_mRNA_mm, refseq_prot, refseq_prot_mm, db, chr, gstart, gend, db_mm, chr_mm,gstart_mm, gend_mm)

local title = "Orthologs"

local entrezTitle = "Entrez"

--check all values to see if exists so don't create nonsense links in infobox must have some alphanumberic character

local entrez_link = ""

if string.match(entrez_gene, '%w+') and entrez_gene ~= "n/a" then

entrez_link = "[http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=gene&cmd=retrieve&dopt=default&list_uids="..entrez_gene.."&rn=1 "..entrez_gene.."]"

end

local entrez_mm_link = entrez_gene_mm

if string.match(entrez_gene_mm, '%w+') and entrez_gene_mm ~= "n/a" then

entrez_mm_link = "[http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=gene&cmd=retrieve&dopt=default&list_uids="..entrez_gene_mm.."&rn=1 "..entrez_gene_mm.."]"

end

local ensemblTitle = "Ensembl"

local ensembl_link = ensembl --default to result so n/a defined

if string.match(ensembl, '%w+') and ensembl ~= "n/a" then

ensembl_link = "[http://www.ensembl.org/Homo_sapiens/geneview?gene="..ensembl..";db=core".." "..ensembl.."]"

end

local ensembl_mm_link = ensembl_mm

if string.match(ensembl_mm, '%w+') and ensembl_mm ~= "n/a" then

ensembl_mm_link = "[http://www.ensembl.org/Mus_musculus/geneview?gene="..ensembl_mm..";db=core".." "..ensembl_mm.."]"

end

--create lists of uniprot ID

local uniprotTitle = "UniProt"

local uniprot_url = "http://www.uniprot.org/uniprot/"

local uniprot_link = "n/a"

local uniprot_collapse

local uniprot_default = ""

--split string and loop through concatenate by

local split_uniprot = mw.text.split(uniprot, ",")

local uniprot_link_list = {}

local hash = {} --storage to look for duplicated values

for k,v in ipairs(split_uniprot) do

if not hash[v] then --only add if not found previously..some encodes uniprotID dup in different encodes

local label = mw.text.trim(v)

local concat_uniprot_link = uniprot_url .. label

if string.match(uniprot, '%w+') and uniprot ~= "n/a" then

uniprot_link_list[#uniprot_link_list+1] = "[" .. concat_uniprot_link .. " " ..label .. "]"

end

hash[v] = true

end

end

--if less than 5 don't create collapsible list

if table.getn(uniprot_link_list) < 5 then

uniprot_collapse = "none"

if uniprot_default == nil and table.getn(uniprot_link_list) == 0 then uniprot_link = "n/a" end

else

uniprot_collapse = "collapsible collapsed"

uniprot_default = table.remove(uniprot_link_list, 1) .. '
' .. table.remove(uniprot_link_list, 1) .. '
' ..table.remove(uniprot_link_list, 1) .. '
' .. table.remove(uniprot_link_list, 1) .. '
' .. table.remove(uniprot_link_list, 1) .. '
'--get first 5 elements in table and use for display

end

if uniprot_link_list[#uniprot_link_list] then

uniprot_link = table.concat(uniprot_link_list, "
")

end

--mouse uniprot lists

local uniprot_mm_link = "n/a"

local uniprot_mm_collapse

local uniprot_mm_default = ""

--split string and loop through concatenate by

local split_uniprot_mm = mw.text.split(uniprot_mm, ",")

local uniprot_mm_link_list = {}

for k,v in ipairs(split_uniprot_mm) do

local label = mw.text.trim(v)

local concat_uniprot_link = uniprot_url .. label

if string.match(uniprot_mm, '%w+') and uniprot_mm ~= "n/a" then

uniprot_mm_link_list[#uniprot_mm_link_list+1] = "[" .. concat_uniprot_link .. " " ..label .. "]"

end

end

--if less than 5 don't create collapsible list

if table.getn(uniprot_mm_link_list) < 5 then

uniprot__mm_collapse = "none"

if uniprot_mm_default == nil and table.getn(uniprot_mm_link_list) == 0 then uniprot_mm_link = "n/a" end

else

uniprot_mm_collapse = "collapsible collapsed"

uniprot_mm_default = table.remove(uniprot_link_mm_list, 1) .. '
' .. table.remove(uniprot_link_mm_list, 1) .. '
' ..table.remove(uniprot_link_mm_list, 1) .. '
' .. table.remove(uniprot_link_mm_list, 1) .. '
' .. table.remove(uniprot_link_mm_list, 1) .. '
'--get first 5 elements in table and use for display

end

if uniprot_mm_link_list[#uniprot_mm_link_list] then

uniprot_mm_link = table.concat(uniprot_mm_link_list, "
")

end

local ncbi_link = "http://www.ncbi.nlm.nih.gov/entrez/viewer.fcgi?val="

local refseq_mRNATitle = "RefSeq (mRNA)"

--create list of links for refSeq mRNA

local refseq_mRNA_link = "n/a"

local refseq_mRNA_collapse

local refseq_mRNA_default = ""

--split string and loop through concatenate by

local split_refseq_mRNA = mw.text.split(refseq_mRNA, ",")

local link_list = {}

for k,v in ipairs(split_refseq_mRNA) do

local label = mw.text.trim(v)

local concat_ncbi_link = ncbi_link .. label

if string.match(v, '%w+') and v ~= "n/a" then

link_list[#link_list+1] = "[" .. concat_ncbi_link .. " " ..label .. "]"

end

end

--if less than 5 don't create collapsible list

if table.getn(link_list) < 6 then

refseq_mRNA_collapse = "none"

if refseq_mRNA_default == nil and table.getn(link_list) == 0 then refseq_mRNA_link = "n/a" end

else

refseq_mRNA_collapse = "collapsible collapsed"

refseq_mRNA_default = table.remove(link_list, 1) .. '
' .. table.remove(link_list, 1) .. '
' ..table.remove(link_list, 1) .. '
' .. table.remove(link_list, 1) .. '
' .. table.remove(link_list, 1) .. '
'--get first 5 elements in table and use for display

end

if link_list[#link_list] then

refseq_mRNA_link = table.concat(link_list, "
")

end

--create list of links for refSeq mRNA for mouse

local refseq_mRNA_mm_link = "n/a"

local refseq_mRNA_mm_collapse

local refseq_mRNA_mm_default = ""

local split_refseq_mRNA_mm = mw.text.split(refseq_mRNA_mm, ",")

local link_list_mm = {}

for k,v in ipairs(split_refseq_mRNA_mm) do

if string.match(v, '%w+') and v ~= "n/a" then

local label = mw.text.trim(v)

local concat_ncbi_link = ncbi_link .. label

if string.match(v, '%w+') and v ~= "n/a" then

link_list_mm[#link_list_mm+1] = "[" .. concat_ncbi_link .. " " ..label .. "]"

end

end

end

--if less than 5 don't create collapsible list

if table.getn(link_list_mm) < 6 then

refseq_mRNA_mm_collapse = "none"

if refseq_mRNA_mm_default == nil and table.getn(link_list_mm) == 0 then refseq_mRNA_mm_link = "n/a" end

else

refseq_mRNA_mm_collapse = "collapsible collapsed"

refseq_mRNA_mm_default = table.remove(link_list_mm, 1) .. '
' .. table.remove(link_list_mm, 1) .. '
' ..table.remove(link_list_mm, 1) .. '
' .. table.remove(link_list_mm, 1) .. '
' .. table.remove(link_list_mm, 1) .. '
'--get first 5 elements in table and use for display

end

if link_list_mm[#link_list_mm] then

refseq_mRNA_mm_link = table.concat(link_list_mm, "
")

end

local refseq_protTitle = "RefSeq (protein)"

--create list of links for human refseq protein

local refseq_prot_link = "n/a"

local refseq_prot_collapse

local refseq_prot_default = ""

local split_refseq_prot = mw.text.split(refseq_prot, ",")

local link_list_prot = {}

for k,v in ipairs(split_refseq_prot) do

if string.match(v, '%w+') and v ~= "n/a" then

local label = mw.text.trim(v)

local concat_ncbi_link = ncbi_link .. label

if string.match(v, '%w+') and v ~= "n/a" then

link_list_prot[#link_list_prot+1] = "[" .. concat_ncbi_link .. " " ..label .. "]"

end

end

end

--if less than 5 don't create collapsible list

if table.getn(link_list_prot) < 6 then

refseq_prot_collapse = "none"

if refseq_prot_default == nil and table.getn(link_list_prot) == 0 then refseq_prot_link = "n/a" end

else

refseq_prot_collapse = "collapsible collapsed"

refseq_prot_default = table.remove(link_list_prot, 1) --get first element in table and use for display

end

if link_list_prot[#link_list_prot] then

refseq_prot_link = table.concat(link_list_prot, "
")

end

--create list of links for mouse refseq protein

local refseq_prot_mm_link = "n/a"

local refseq_prot_mm_collapse

local refseq_prot_mm_default = ""

local split_refseq_prot_mm = mw.text.split(refseq_prot_mm, ",")

local link_list_mm = {}

for k,v in ipairs(split_refseq_prot_mm) do

if string.match(v, '%w+') and v ~= "n/a" then

local label = mw.text.trim(v)

local concat_ncbi_link = ncbi_link .. label

if string.match(v, '%w+') and v ~= "n/a" then

link_list_mm[#link_list_mm+1] = "[" .. concat_ncbi_link .. " " ..label .. "]"

end

end

end

--if less than 5 don't create collapsible list

if table.getn(link_list_mm) < 6 then

refseq_prot_mm_collapse = "none"

if refseq_prot_mm_default == nil and table.getn(link_list_mm) == 0 then refseq_prot_mm_link = "n/a" end

else

refseq_prot_mm_collapse = "collapsible collapsed"

refseq_prot_mm_default = table.remove(link_list_mm, 1) --get first element in table and use for display

end

if link_list_mm[#link_list_mm] then

refseq_prot_mm_link = table.concat(link_list_mm, "
")

end

local locTitle = "Location (UCSC)"

local gstart_mb = p.locToMb(gstart, 2)

local gend_mb = p.locToMb(gend, 2)

local chr_loc_link = ""

if (string.match(db, '%w+') and string.match(chr, '%w+') and string.match(gstart, '%w+') and string.match(gend, '%w+') )then

chr_loc_link = "[http://genome.ucsc.edu/cgi-bin/hgTracks?org=Human&db="..db.."&position=chr"..chr..":"..gstart.."-"..gend.." ".."Chr "..chr..": "..gstart_mb.." – "..gend_mb.." Mb]"

else

chr_loc_link = "n/a"

end

local gstart_mm_mb = p.locToMb(gstart_mm, 2)

local gend_mm_mb = p.locToMb(gend_mm, 2)

local chr_loc_mm_link = ""

if (string.match(db_mm, '%w+') and string.match(chr_mm, '%w+') and string.match(gstart_mm, '%w+') and string.match(gend_mm, '%w+') )then

chr_loc_mm_link = "[http://genome.ucsc.edu/cgi-bin/hgTracks?org=Human&db="..db_mm.."&position=chr"..chr_mm..":"..gstart_mm.."-"..gend_mm.." ".."Chr "..chr_mm..": "..gstart_mm_mb.." – "..gend_mm_mb.." Mb]"

else

chr_loc_link = "n/a"

end

local pubmedTitle = "PubMed search"

local pubmed_link = entrez_gene

if string.match(entrez_gene, '%w+') and entrez_gene ~= "n/a" then

pubmed_link = "[http://www.ncbi.nlm.nih.gov/sites/entrez?db=gene&cmd=Link&LinkName=gene_pubmed&from_uid="..entrez_gene.."]"

end

local pubmed_mm_link = entrez_gene_mm

if string.match(entrez_gene_mm, '%w+') and entrez_gene_mm ~= "n/a" then

pubmed_mm_link = "[http://www.ncbi.nlm.nih.gov/sites/entrez?db=gene&cmd=Link&LinkName=gene_pubmed&from_uid="..entrez_gene_mm.."]"

end

root

:tag('tr')

:tag('th')

:attr('colspan', '4')

:css('text-align', 'center')

:css('background-color', titleBGcolor)

:wikitext(title)

:done()

:newline()

:tag('tr')

:tag('th')

:attr('scope', 'row')

:css('background-color', sideTitleBGcolor)

:wikitext("Species")

:done()

:newline()

:tag('td')

:wikitext("Human")

:done()

:newline()

:tag('td')

:wikitext("Mouse")

:done()

:newline()

:done()

---this section would be nice to do in a loop if the tag closures were figured out

:tag('tr')

:tag('th')

:attr('scope', 'row')

:css('background-color', sideTitleBGcolor)

:wikitext(entrezTitle)

:done()

:newline()

:tag('td')

:tag('span')

:attr('class', 'plainlinks')

:wikitext(entrez_link)

:done()

:newline()

:tag('td')

:tag('span')

:attr('class', 'plainlinks')

:wikitext(entrez_mm_link)

:done()

:newline()

:tag('tr')

:tag('th')

:attr('scope', 'row')

:css('background-color', sideTitleBGcolor)

:wikitext(ensemblTitle)

:done()

:newline()

:tag('td')

:tag('span')

:attr('class', 'plainlinks')

:wikitext(ensembl_link)

:done()

:newline()

:tag('td')

:tag('span')

:attr('class', 'plainlinks')

:wikitext(ensembl_mm_link)

:done()

:newline()

:tag('tr')

:tag('th')

:attr('scope', 'row')

:css('background-color', sideTitleBGcolor)

:wikitext(uniprotTitle)

:done()

:newline()

:tag('td')

:tag('table')

:attr('class', uniprot_collapse)

:css('padding', '0')

:css('border', 'none')

:css('margin', '0')

:css('width', '100%')

:css('text-align', 'right')

:tag('tr')

:newline()

:tag('th')

:attr('colspan', '1')

:tag('span')

:attr('class', 'plainlinks')

:wikitext(uniprot_default)

:done()

:done()

:newline()

:done()

:tag('tr')

:tag('td')

:attr('colspan', '1')

:newline()

:tag('p')

:attr('class', 'plainlinks')

:wikitext(uniprot_link)

:newline()

:done()

:done()

:done()

:done()

:newline()

:tag('td')

:tag('table')

:attr('class', uniprot_mm_collapse)

:css('padding', '0')

:css('border', 'none')

:css('margin', '0')

:css('width', '100%')

:css('text-align', 'right')

:tag('tr')

:newline()

:tag('th')

:attr('colspan', '1')

:tag('span')

:attr('class', 'plainlinks')

:wikitext(uniprot_mm_default)

:done()

:done()

:newline()

:done()

:tag('tr')

:tag('td')

:attr('colspan', '1')

:newline()

:tag('p')

:attr('class', 'plainlinks')

:wikitext(uniprot_mm_link)

:newline()

:done()

:done()

:done()

:done()

:newline()

:tag('tr')

:tag('th')

:attr('scope', 'row')

:css('background-color', sideTitleBGcolor)

:wikitext(refseq_mRNATitle)

:done()

:newline()

:tag('td') --RNASeq mRNA collapsible table

:tag('table')

:attr('class', refseq_mRNA_collapse)

:css('padding', '0')

:css('border', 'none')

:css('margin', '0')

:css('width', '100%')

:css('text-align', 'right')

:tag('tr')

:newline()

:tag('th')

:attr('colspan', '1')

:attr('class', 'plainlinks')

:wikitext(refseq_mRNA_default)

:done()

:newline()

:done()

:tag('tr')

:tag('td')

:attr('colspan', '1')

:newline()

:tag('p')

:tag('span')

:attr('class', 'plainlinks')

:wikitext(refseq_mRNA_link)

:done()

:newline()

:done()

:done()

:done()

:done()

:tag('td') --RNASeq mRNA collapsible table for mouse

:tag('table')

:attr('class', refseq_mRNA_mm_collapse)

:css('padding', '0')

:css('border', 'none')

:css('margin', '0')

:css('width', '100%')

:css('text-align', 'right')

:tag('tr')

:newline()

:tag('th')

:attr('colspan', '1')

:attr('class', 'plainlinks')

:wikitext(refseq_mRNA_mm_default)

:done()

:newline()

:done()

:tag('tr')

:tag('td')

:attr('colspan', '1')

:newline()

:tag('p')

:tag('span')

:attr('class', 'plainlinks')

:wikitext(refseq_mRNA_mm_link)

:done()

:newline()

:done()

:done()

:done()

:done()

:tag('tr')

:tag('th')

:attr('scope', 'row')

:css('background-color', sideTitleBGcolor)

:wikitext(refseq_protTitle)

:done()

:newline()

:tag('td') --RNASeq protein collapsible table

:tag('table')

:attr('class', refseq_prot_collapse)

:css('padding', '0')

:css('border', 'none')

:css('margin', '0')

:css('width', '100%')

:css('text-align', 'right')

:tag('tr')

:newline()

:tag('th')

:attr('colspan', '1')

:attr('class', 'plainlinks')

:wikitext(refseq_prot_default)

:done()

:newline()

:done()

:tag('tr')

:tag('td')

:attr('colspan', '1')

:newline()

:tag('p')

:tag('span')

:attr('class', 'plainlinks')

:wikitext(refseq_prot_link)

:done()

:newline()

:done()

:done()

:done()

:done()

:tag('td') --RNASeq protein collapsible table for mouse

:tag('table')

:attr('class', refseq_prot_mm_collapse)

:css('padding', '0')

:css('border', 'none')

:css('margin', '0')

:css('width', '100%')

:css('text-align', 'right')

:tag('tr')

:newline()

:tag('th')

:attr('colspan', '1')

:attr('class', 'plainlinks')

:wikitext(refseq_prot_mm_default)

:done()

:done()

:tag('tr')

:tag('td')

:attr('colspan', '1')

:newline()

:tag('p')

:tag('span')

:attr('class', 'plainlinks')

:wikitext(refseq_prot_mm_link)

:done()

:newline()

:done()

:done()

:done()

:done()

:tag('tr')

:tag('th')

:attr('scope', 'row')

:css('background-color', sideTitleBGcolor)

:wikitext(locTitle)

:done()

:newline()

:tag('td')

:tag('span')

:attr('class', 'plainlinks')

:wikitext(chr_loc_link)

:done()

:newline()

:tag('td')

:tag('span')

:attr('class', 'plainlinks')

:wikitext(chr_loc_mm_link)

:done()

:newline()

:tag('tr')

:tag('th')

:attr('scope', 'row')

:css('background-color', sideTitleBGcolor)

:wikitext(pubmedTitle)

:done()

:newline()

:tag('td')

:tag('span')

:attr('class', 'plainlinks')

:wikitext(pubmed_link)

:done()

:newline()

:tag('td')

:tag('span')

:attr('class', 'plainlinks')

:wikitext(pubmed_mm_link)

:done()

:newline()

end

p.formatRow = function(title)

root

:tag('tr')

:tag('td')

:attr('colspan', '4')

:css('text-align', 'center')

:css('background-color', rowBGcolor)

:newline()

--p.createTable('sub')

:tag('table')

:css('padding', '0')

:css('border', 'none')

:css('margin', '0')

:css('width', '100%')

:css('text-align', 'left')

:newline()

:tag('tr') --create title header

:css('background-color',titleBGcolor)

:css('text-align', 'center')

:tag('th')

:attr('colspan',"2")

:wikitext(title)

:done()

:done()

:newline()

end

p.renderFooter = function(Qid, Qid_mm)

local text = "Wikidata"

local hs_link = "View/Edit Human"

local mm_link = ""

local link_no_hs

local link_no_mm

if Qid_mm == "" then

link_no_mm = 0

link_no_hs = 4

else

link_no_mm = 2

link_no_hs = 2

mm_link = "View/Edit Mouse"

end

root

:tag('tr')

:tag('td')

:attr('colspan', '4')

:css('text-align', 'center')

:css('font-size','x-small')

:css('background-color', rowBGcolor)

:wikitext(text)

:done()

:newline()

:tag('table')

:css('padding', '0')

:css('border', 'none')

:css('margin', '0')

:css('width', '100%')

:css('text-align', 'center')

:tag('tr')

:tag('td')

:attr('colspan', link_no_hs)

:css('background-color', rowBGcolor)

:css('text-align', 'center')

:css('font-size','x-small')

:wikitext(hs_link)

:done()

:tag('td')

:attr('colspan', link_no_mm)

:css('background-color', rowBGcolor)

:css('text-align', 'center')

:css('font-size','x-small')

:wikitext(mm_link)

:done()

:newline()

:done()

end

--this code isn't used was hoping could do some generalization of rows

p.rowLabel=function(label)

root

:tag('tr')

:tag('th')

:attr('rowspan', '2')

:css('background-color', sideTitleBGcolor)

:css('width', '43px')

:wikitext(label)

--:done()

end

-- look into entity object

p.getLabel = function(entity)

local data = entity

local f = {'labels','en','value'}

local i = 1

while true do

local index = f[i]

if not index then

if type(data) == "table" then

return mw.text.jsonEncode(data, mw.text.JSON_PRESERVE_KEYS + mw.text.JSON_PRETTY)

else

return tostring(data)

end

end

data = data[index] or data[tonumber(index)]

if not data then

return

end

i = i + 1

end

end

--general function to get value given an entity and property

p.getValue = function(entity, propertyID, return_val)

local claims

if return_val == nil then return_val = "" end

local sep = " " --could ad as input parameter if need be

if entity and entity.claims then

claims = entity.claims[propertyID]

end

if claims then

-- if wiki-linked value output as link if possible

if (claims[1] and claims[1].mainsnak.snaktype == "value" and claims[1].mainsnak.datavalue.type == "wikibase-entityid") then

local out = {}

for k, v in pairs(claims) do

local datav = mw.wikibase.label("Q" .. v.mainsnak.datavalue.value["numeric-id"])

if datav == nil then datav = " " end

out[#out + 1] = datav

end

return table.concat(out, sep)

else

-- just return best values

return entity:formatPropertyValues(propertyID).value

end

else

return return_val

end

end

p.getValueProtein = function(protein_entities, propertyID, return_val)

if return_val == nil then return_val = "" end

local sep = ","

local overall_results = {} --should return empty if nothing assigned

for key, val in pairs(protein_entities) do --in cases where there are multiple encodes we loop through each and return concatenated data as a whole

local claims

local entity = val --each protein in encodes

if entity and entity.claims then

claims = entity.claims[propertyID]

end

if claims then

local results

-- if wiki-linked value output as link if possible

if (claims[1] and claims[1].mainsnak.snaktype == "value" and claims[1].mainsnak.datavalue.type == "wikibase-entityid") then

local out = {}

for k, v in pairs(claims) do

local datav = mw.wikibase.label("Q" .. v.mainsnak.datavalue.value["numeric-id"])

if datav == nil then datav = " " end

out[#out + 1] = datav

end

results = table.concat(out, sep)

else

results = entity:formatPropertyValues(propertyID).value

end

overall_results[#overall_results+1] = results --individual propertyID value stored in this index

end

end

local str_overall_results = table.concat(overall_results, sep) --weirdness happens when add a sep = " " otherwise each value represented one time

if string.match(str_overall_results, '%w+') then

return str_overall_results

else

return return_val

end

end

--general function to get value given an entity and property

p.getQid = function(entity)

local Qid

if entity and entity.id then

Qid = entity.id

return Qid

else

return ""

end

end

--get random value that is preferred ranked

p.getRefseq_mRNA = function(entity, propertyID, return_val)

if return_val == nil then return_val = "" end

local input_rank = "RANK_PREFERRED" ---this is mostly like won't do anything because ranking isn't maintained in wikidata

local claims

if entity.claims then

claims = entity.claims[propertyID]

end

if claims then

-- if wiki-linked value output as link if possible

if (claims[1] and claims[1].mainsnak.snaktype == "value" and claims[1].mainsnak.datavalue.type == "wikibase-entityid" ) then

local out = {}

for k, v in pairs(claims) do

local sitelink = mw.wikibase.sitelink("Q" .. v.mainsnak.datavalue.value["numeric-id"])

local label = mw.wikibase.label("Q" .. v.mainsnak.datavalue.value["numeric-id"])

if label == nil then label = "Q" .. v.mainsnak.datavalue.value["numeric-id"] end

if sitelink then

out[#out + 1] = "" .. label .. ""

else

out[#out + 1] = "" .. label .. ""

end

end

return table.concat(out, ", ")

else

local results = entity:formatPropertyValues(propertyID, mw.wikibase.entity.claimRanks).value

--loop through results until get a NP or NM or just return whatever is in first element

--[[local results_split = mw.text.split(results, ",")

local preffered_results = " "

if results_split[1] then

preferred_result = mw.text.trim(results_split[1]) --return first element if desired prefix not found and remove whitespace

end

local id --refseq id in question

for i, id in ipairs(results_split) do

local trim_id = mw.text.trim(id)

if string.match( trim_id, '^NM_%d+') then

preferred_result = trim_id --overwrite each time found only need one to display

end

end

if preferred_result then

return preferred_result --return a id starting with NP or NM

else

return return_val --return first element because desired prefix not found and remove whitespaces

end

--]]

return results

end

else

return return_val

end

end

p.getRefseq_protein = function(protein_entities, propertyID, return_val)

local sep = ","

local overall_results = {} --should return empty if nothing assigned

for key, val in pairs(protein_entities) do --in cases where there are multiple encodes we loop through each and return concatenated data as a whole

local claims

local entity = val --each protein in encodes

if entity.claims then

claims = entity.claims["P637"]

end

if claims then

local results

-- if wiki-linked value output as link if possible

if (claims[1] and claims[1].mainsnak.snaktype == "value" and claims[1].mainsnak.datavalue.type == "wikibase-entityid" ) then

local out = {}

for k, v in pairs(claims) do

local datav = mw.wikibase.label("Q" .. v.mainsnak.datavalue.value["numeric-id"])

if datav == nil then datav = " " end

out[#out + 1] = datav

end

results = table.concat(out, sep)

else

results = entity:formatPropertyValues("P637", mw.wikibase.entity.claimRanks).value

end

overall_results[#overall_results+1] = results --a list is in each index

end

end

--why are there duplicate results here

local str_overall_results = table.concat(overall_results, sep)

return str_overall_results

end

--[[

local results_split = mw.text.split(str_overall_results, sep) --split complete list so can loop through..probably a more direct way to do this

--loop through results until get a NP or NM or just return whatever is in first element

local preffered_result = results_split[1] or ""

for i, id in ipairs(results_split) do

local trim_id = mw.text.trim(id)

--check of id starts with NP or NM

if string.match( trim_id, '^NP_%d+') then

preferred_result = trim_id --overwrite each time found only need one to display

end

end

--check if something in preffered_result if not get first element in result_split

if p.isempty(preffered_result) then

return return_val

else

return preferred_result --return a id starting with NP or NM

end

end --]]

--gets an image

p.getImage = function(entity, propertyID, sep, imgsize)

local claims

if entity and entity.claims then

claims = entity.claims[propertyID]

end

if claims then

if (claims[1] and claims[1].mainsnak.datatype == "commonsMedia") then

local out = {}

for k, v in pairs(claims) do

local filename = v.mainsnak.datavalue.value

out[#out + 1] = "" .. imgsize .. ""

end

return table.concat(out, sep)

else

return ""

end

else

return ""

end

end

p.getPDB = function(protein_entities)

local pdb_propertyID = "P638"

local overall_results = {}

for key, val in pairs(protein_entities) do --in cases where there are multiple encodes we loop through each and return concatenated data as a whole

local claims

local entity = val

if entity and entity.claims then

claims = entity.claims[pdb_propertyID]

end

local sitelink = "http://www.rcsb.org/pdb/explore/explore.do?pdbId="

if claims then

local results

if (claims[1] and claims[1].mainsnak.snaktype == "value") then

local out = {}

for k, v in pairs(claims) do

local label = mw.wikibase.label(v.mainsnak.datavalue.value)

if label == nil then label = v.mainsnak.datavalue.value end

if sitelink then

out[#out + 1] = "[" .. sitelink .. label .. " " ..label .. "]"

else

out[#out + 1] = "" .. label .. ""

end

end

results = table.concat(out, ", ")

else

results = entity:formatPropertyValues(propertyID, mw.wikibase.entity.claimRanks).value

end

overall_results[#overall_results+1] = results --individual propertyID values stored in this index

end

end

return table.concat(overall_results, ",%%s")

end

function p.getAliases(entity)

a = ''

if entity['aliases'] ~= nil then

local test = entity['aliases']['en']

if test then

for key, value in ipairs(test) do

a = a .. ', ' .. value['value']

end

return a

else

return ""

end

else

return ""

end

end

--get a geneome start P644 or end P645

p.getChromosomeLoc = function(entity, propertyID, prefix)

-- will contain the numeric value for the requested coordinate

local output = ""

local sep = " "

-- can only be P644 (genomic start) or P645 (genomic end) for this to work

-- should probably try to catch that. Might also increase legibility to use specific variable names when possible

-- local propertyID = mw.text.trim(frame.args[1] or "")

-- this can really only be P659 right now. I'm not sure of the value of including it as a parameter as other values will likely break this function

local qualifierID = "P659" --mw.text.trim(frame.args[2] or "")

-- Why do we include this here? What should happen if FETCH_WIKIDATA is not included?

--local input_parm = mw.text.trim(frame.args[3] or "")

-- this can needs to be fed to the function either by a call to {{#invoke:Wikidata|pageId}} or by setting it directly (e.g. if the function was applied on a page other than the targeted gene)

--alert if this id is not a valid thing in wikidata, a Lua error will occur that says

--The ID entered is unknown to the system. Please use a valid entity ID.

--local itemID = mw.text.trim(frame.args[4] or "")

-- will track the different builds pulled from the qualifiers

local newest_build = "0"

-- starts the process

--local entity = mw.wikibase.getEntityObject(itemID)

local claims

--gets a table of claims on the (genomic start or end) property Q19847637

if entity and entity.claims then

claims = entity.claims[propertyID]

end

--will return nothing if no claims are found

if claims then

--checking to be sure claims is populated, not sure it its needed

if (claims[1] ) then

--useful for debugging

--local out = {}

--pulls the genome location from the claim

for k, v in pairs(claims) do

local location = v.mainsnak.datavalue.value

--debugging

--out[#out + 1] = k.." location:" .. location.. " || "

--gets the qualifiers linked to the current claim

local quals

if v.qualifiers then

quals = v.qualifiers.P659

end

--if there are any

if quals then

for qk, qv in pairs(quals) do

local qual_obj_id = "Q"..qv.datavalue.value["numeric-id"]

--get to the entity targeted by the qualifier property. Genome builds are Items in wikidata

local qual_obj = mw.wikibase.getEntityObject(qual_obj_id)

local alias = ""

--this uses the aliases to pull out version numbers

--seems like there ought to be a better way to do this, but likely would need to change the data added by the bot

if qual_obj["aliases"] ~= nil then

local test = qual_obj["aliases"]["en"]

for key, value in ipairs(test) do

if string.match(value['value'], prefix) then

alias = value['value']

local build_no = alias:gsub(prefix,"")

--report only the most location associated with the most recent build

--if there is more than one location per build, just give one back as that is not our problem right now.

if build_no > newest_build then

output = location

newest_build = build_no

end

end

end

end

end

--in case there are no qualifiers, but there is a location, might as well return it

else output = location

end

end

return output

else

return ""

end

else

return ""

--debug

--"no claims for "..itemID.." prop "..propertyID

end

end

p.getAliasFromGenomeAssembly = function(entity, prefix)

-- will contain the numeric value for the requested coordinate

local output = ""

local sep = " "

local propertyID = "P644" --genomic start used

local qualifierID = "P659"

local newest_build = "0"

local claims

if entity.claims then

claims = entity.claims[propertyID]

end

--will return nothing if no claims are found

if claims then

--checking to be sure claims is populated, not sure it its needed

if (claims[1] ) then

--useful for debugging

--local out = {}

--pulls the genome location from the claim

for k, v in pairs(claims) do

local quals

if v.qualifiers then

quals = v.qualifiers.P659

end

--if there are any

if quals then

for qk, qv in pairs(quals) do

local qual_obj_id = "Q"..qv.datavalue.value["numeric-id"]

--get to the entity targeted by the qualifier property. Genome builds are Items in wikidata

local qual_obj = mw.wikibase.getEntityObject(qual_obj_id)

local alias = ""

--this uses the aliases to pull out version numbers

--seems like there ought to be a better way to do this, but likely would need to change the data added by the bot

if qual_obj["aliases"] ~= nil then

local test = qual_obj["aliases"]["en"]

for key, value in ipairs(test) do

if string.match(value['value'], prefix) then

alias = value['value']

local build_no = alias:gsub(prefix,"")

--report only the most location associated with the most recent build

--if there is more than one location per build, just give one back as that is not our problem right now.

if build_no > newest_build then

newest_build = build_no

end

end

end

end

end

--in case there are no qualifiers, but there is a location, might as well return it

else output = location

end

end

return prefix..newest_build

else

return ""

end

else

return ""

end

end

p.trimChromosome = function(entity)

local string_to_trim = p.getValue(entity, "P1057")

local out = ''

if string.find(string_to_trim, 'chromosome.') then

out = string_to_trim:gsub("chromosome ", "")

end

if string.find(string_to_trim, 'mouse.chromosome.') then

out = string_to_trim:gsub("mouse chromosome ", "")

end

return out

end

p.locToMb = function(num, idp)

num = tonumber(num)

if num == nil then

return ""

else

local mb = num/1000000

local mult = 10^(idp or 0)

return math.floor(mb * mult + 0.5) / mult

end

end

p.isempty = function(s)

return s == nil or s == ''

end

p.getGO = function(protein_entities, propertyID)

--propertyID ie molecular, cellular, function

local overall_results = {}

local results = "" --string to return

for key, val in pairs(protein_entities) do

local claims

local entity = val

if entity.claims then

claims = entity.claims[propertyID] -- ie molecular, cellular, function

end

local propertyID_child = "P686" -- Gene Ontology ID

if claims then

if (claims[1] and claims[1].mainsnak.snaktype == "value" and claims[1].mainsnak.datavalue.type == "wikibase-entityid") then

--local out = {}

for k, v in pairs(claims) do

local itemID_child = "Q" .. v.mainsnak.datavalue.value["numeric-id"] --get Qid of property item so can get the GOid

local entity = mw.wikibase.getEntityObject(itemID_child)

local claims

local result_GOID = ''

if entity and entity.claims then claims = entity.claims[propertyID_child] end

if claims then

result_GOID = entity:formatPropertyValues(propertyID_child, mw.wikibase.entity.claimRanks).value

else

result_GOID = nil --no GO ID

end

local sitelink = "http://amigo.geneontology.org/amigo/term/"

local label = mw.wikibase.label("Q" .. v.mainsnak.datavalue.value["numeric-id"])

if label == nil then label = "Q" .. v.mainsnak.datavalue.value["numeric-id"] end

local wiki_link = ""

if sitelink and result_GOID ~= nil then

wiki_link = "• [" .. sitelink .. result_GOID .. " " .. label .."]
"

else

wiki_link = "• " .. label .. "
"

end

overall_results[#overall_results+1] = wiki_link

end

else

results = entity:formatPropertyValues(propertyID, mw.wikibase.entity.claimRanks).value

end

end

--overall_results[#overall_results+1] = results --each protein GO terms stored in this index, so table contains all the GO terms with duplicates

end

local hash = {} --temp check

local res = {} --no dups

for _,v in ipairs(overall_results) do

if (not hash[v]) then

res[#res+1] = v

hash[v] = true

end

end

return table.concat(res, "")

end

return p