# HG changeset patch
# User immport-devteam
# Date 1488218599 18000
# Node ID 8547aedf1350120dbf31981a1d6894d8ac06d86b
Uploaded
diff -r 000000000000 -r 8547aedf1350 profile_cl/getOntology.R
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/profile_cl/getOntology.R Mon Feb 27 13:03:19 2017 -0500
@@ -0,0 +1,28 @@
+# Cell Ontology Module for Galaxy
+# FlowCL
+######################################################################
+# Copyright (c) 2016 Northrop Grumman.
+# All rights reserved.
+######################################################################
+#
+# Version 1
+# Cristel Thomas
+#
+#
+
+library(flowCL)
+library(base)
+
+getOntology <- function(output_file, markers) {
+ res <- flowCL(markers, ResetArch = TRUE)
+ if (length(res) == 6) {
+ report <- capture.output(res$Table)
+ sink(output_file)
+ cat(report, sep = "\n")
+ sink()
+ }
+}
+
+args <- commandArgs(trailingOnly = TRUE)
+markers <- paste(args[3:length(args)], collapse="")
+getOntology(args[2], markers)
diff -r 000000000000 -r 8547aedf1350 profile_cl/profileCLs.py
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/profile_cl/profileCLs.py Mon Feb 27 13:03:19 2017 -0500
@@ -0,0 +1,163 @@
+#!/usr/bin/env python
+
+######################################################################
+# Copyright (c) 2016 Northrop Grumman.
+# All rights reserved.
+######################################################################
+from __future__ import print_function
+import sys
+import os
+from argparse import ArgumentParser
+from jinja2 import Environment, FileSystemLoader
+
+profile_key = {
+ "1": "-",
+ "2": "lo",
+ "3": "+",
+ "4": "hi"
+}
+
+
+def run_flowCL(phenotype, output_txt, output_pdf, tool):
+ run_command = " ". join(["Rscript --slave --vanilla", tool, "--args", output_txt, phenotype])
+ os.system(run_command)
+ get_graph = " ".join(["mv flowCL_results/*.pdf", output_pdf])
+ os.system(get_graph)
+ return
+
+
+def generate_flowCL_query(list_markers, list_types):
+ if (len(list_markers) != len(list_types)):
+ return("pb with headers")
+ query = []
+ # go through both lists, remove fsc/ssc
+ for i in range(1, len(list_markers)):
+ if not list_markers[i].startswith("FSC") and not list_markers[i].startswith("SSC"):
+ query.append(list_markers[i].upper())
+ query.append(profile_key[list_types[i]])
+ # return concatenated string
+ return("".join(query))
+
+
+def translate_profiles(input_file, tool_dir, output, html_dir):
+ os.mkdir(html_dir)
+
+ tool = "/".join([tool_dir, "getOntology.R"])
+ html_table = "".join([html_dir, "/CLprofiles.txt"])
+ score_table = "".join(["cp ", input_file, " ", html_dir, "/scores.txt"])
+ os.system(score_table)
+
+ # read profile
+ with open(input_file, "r") as flock_profiles, open(html_table, "w") as out:
+ headers = flock_profiles.readline()
+ headers = headers.strip()
+ # get all headers except for last 2 (count + percentage)
+ markers = headers.split("\t")[:-2]
+ counter = 0
+
+ out.write("Population\tFlowCL Query\tNb Results\tLink to PDF\t")
+ out.write("Top Result Label\tTop Result Score\tTop Result CL\n")
+ queries = {}
+ # create marker query for each population
+ for lines in flock_profiles:
+ lines = lines.strip("\n")
+ pop_profile = lines.split("\t")[:-2]
+ flowcl_query = generate_flowCL_query(markers, pop_profile)
+ counter += 1
+ nb_results = "0"
+ top_label = "no_match"
+ top_score = "NA"
+ top_CL = "NA"
+ pdf_link = "NA"
+ # check if query was run before
+ if flowcl_query not in queries:
+ # create filenames for results & graphs
+ txt = "".join(["flowcl_pop", str(counter).zfill(2), ".txt"])
+ text_result = "/".join([html_dir, txt])
+ graph = "".join(["flowcl_pop", str(counter).zfill(2), ".pdf"])
+ graph_output = "/".join([html_dir, graph])
+ # run flowCL for each marker profile
+ run_flowCL(flowcl_query, text_result, graph_output, tool)
+
+ # test that text file exists if not results are all NAs:
+ if os.path.isfile(text_result):
+ with open(text_result, "r") as res:
+ for line in res:
+ if line.startswith("Score"):
+ data = line.split(") ")
+ top_score = data[2][:-2]
+ tot_results = len(data) - 2
+ nb_results = str(tot_results)
+ if tot_results == 5:
+ if len(data[6].split("+")) > 1:
+ nb_results = "5+"
+ elif line.startswith("Cell ID"):
+ prep_link = line.split(") ")[1][:-2]
+ cl = prep_link.replace("_", ":")
+ link = "".join([''])
+ top_CL = "".join([link, prep_link, ""])
+ elif line.startswith("Cell Label"):
+ top_label = line.split(") ")[1][:-2]
+ pdf_link = "".join(['PDF'])
+ tmpflowcl_query = "".join(['', flowcl_query, ''])
+
+ queries[flowcl_query] = {
+ "query": tmpflowcl_query,
+ "results": nb_results,
+ "pdf": pdf_link,
+ "label": top_label,
+ "score": top_score,
+ "CL": top_CL
+ }
+ # write query results to CLprofiles.txt
+ out.write("\t".join([pop_profile[0],
+ queries[flowcl_query]["query"],
+ queries[flowcl_query]["results"],
+ queries[flowcl_query]["pdf"],
+ queries[flowcl_query]["label"],
+ queries[flowcl_query]["score"],
+ queries[flowcl_query]["CL"]]) + "\n")
+
+ env = Environment(loader=FileSystemLoader(tool_dir + "/templates"))
+ template = env.get_template("profileCLs.template")
+
+ real_directory = html_dir.replace("/job_working_directory", "")
+ context = {'outputDirectory': real_directory}
+ overview = template.render(**context)
+ with open(output, "w") as outf:
+ outf.write(overview)
+
+
+if __name__ == "__main__":
+ parser = ArgumentParser(
+ prog="getCLs_from_profile",
+ description="runs flowCL on a each population defined by FLOCK.")
+
+ parser.add_argument(
+ '-i',
+ dest="input_file",
+ required=True,
+ help="File location for the profile.txt from FLOCK.")
+
+ parser.add_argument(
+ '-o',
+ dest="output",
+ required=True,
+ help="Name of the output html file.")
+
+ parser.add_argument(
+ '-d',
+ dest="html_dir",
+ required=True,
+ help="Path to html supporting directory.")
+
+ parser.add_argument(
+ '-t',
+ dest="tool_dir",
+ required=True,
+ help="Path to the tool directory")
+
+ args = parser.parse_args()
+
+ translate_profiles(args.input_file, args.tool_dir, args.output, args.html_dir)
+ sys.exit(0)
diff -r 000000000000 -r 8547aedf1350 profile_cl/profileCLs.xml
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/profile_cl/profileCLs.xml Mon Feb 27 13:03:19 2017 -0500
@@ -0,0 +1,116 @@
+
+ with cell ontology
+
+ jinja2
+ r
+ bioconductor-flowcl
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+ 10.1093/bioinformatics/btu807
+
+
diff -r 000000000000 -r 8547aedf1350 profile_cl/static/flowtools/css/flowCL.css
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/profile_cl/static/flowtools/css/flowCL.css Mon Feb 27 13:03:19 2017 -0500
@@ -0,0 +1,34 @@
+/*
+** DataTables config
+*/
+div.dt-buttons {
+ float: right;
+}
+th {
+ font-size:16px;
+ text-align: left;
+}
+body td {
+ max-width: 50px;
+ font-size: 14px;
+ overflow: hidden;
+ text-overflow: ellipsis;
+}
+.cltable thead {
+ display:none;
+}
+.cltable td {
+ max-width: 40px;
+ word-break: break-all;
+ word-wrap: break-word;
+}
+.firstcol {
+ width: 160px;
+}
+.smallcols {
+ width: 100px;
+}
+div.dataTables_wrapper {
+ width: 95%;
+ margin: 0 auto;
+}
diff -r 000000000000 -r 8547aedf1350 profile_cl/static/flowtools/js/profileCLs.js
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/profile_cl/static/flowtools/js/profileCLs.js Mon Feb 27 13:03:19 2017 -0500
@@ -0,0 +1,85 @@
+// Copyright (c) 2016 Northrop Grumman.
+// All rights reserved.
+
+var cl_table = './CLprofiles.txt';
+var scores_table = './scores.txt';
+
+var displayCLTable = function(){
+ d3.text(cl_table, function(error, data){
+ var clHdgs = [],
+ clHTML = '',
+ clTableData = [],
+ clHeadings = [];
+
+ if (error){
+ alert("Problem retrieving data");
+ return;
+ }
+ clHdgs = data.split("\n")[0].split("\t");
+ data = d3.tsv.parse(data);
+ clTableData = $.extend(true, [], data);
+
+ clHdgs.forEach(function(d,i){
+ clHeadings.push({"data" : d, "title" : d});
+ });
+
+ $('#clprofiles').html(clHTML);
+ var clTable = $('#cltable').DataTable({
+ columns: clHeadings,
+ dom: '<"top"Bi>t<"bottom"lp><"clear">',
+ pageLength: 25,
+ order: [[ 0, "asc" ]],
+ data: clTableData,
+ buttons: [
+ 'copy', 'pdfHtml5','csvHtml5'
+ ],
+ columnDefs: [
+ {
+ targets: [0,2,3],
+ className: "smallcols"
+ },
+ {
+ targets: 4,
+ className: "dt-body-left"
+ },
+ {
+ targets: [5,6],
+ className: "firstcol"
+ }]
+ });
+ });
+};
+
+var displayScoresTable = function(){
+ d3.text(scores_table, function(error, data){
+ var scoreHTML = '',
+ scoreHdgs = [],
+ scoreTableData = [],
+ scoreHeadings = [];
+ if (error){
+ alert("Problem retrieving data");
+ return;
+ }
+ scoreHdgs = data.split("\n")[0].split("\t");
+ data = d3.tsv.parse(data);
+
+ $('#scores').html(scoreHTML);
+
+ scoreTableData = $.extend(true, [], data);
+
+ scoreHdgs.forEach(function(d,i){
+ scoreHeadings.push({"data" : d, "title" : d});
+ });
+
+ var scoreTable = $('#scoretable').DataTable({
+ columns: scoreHeadings,
+ pageLength: 25,
+ order: [[ 0, "asc" ]],
+ dom: '<"top"Bi>t<"bottom"lp><"clear">',
+ data: scoreTableData,
+ buttons: [
+ 'copy', 'pdfHtml5','csvHtml5'
+ ],
+ });
+ });
+};
diff -r 000000000000 -r 8547aedf1350 profile_cl/static/images/flowtools/popprofiles.png
Binary file profile_cl/static/images/flowtools/popprofiles.png has changed
diff -r 000000000000 -r 8547aedf1350 profile_cl/static/images/flowtools/profileCL.png
Binary file profile_cl/static/images/flowtools/profileCL.png has changed
diff -r 000000000000 -r 8547aedf1350 profile_cl/templates/profileCLs.template
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/profile_cl/templates/profileCLs.template Mon Feb 27 13:03:19 2017 -0500
@@ -0,0 +1,91 @@
+
+
+
+FLOCK Populations Ontology
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
diff -r 000000000000 -r 8547aedf1350 profile_cl/test-data/CLprofiles.txt
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/profile_cl/test-data/CLprofiles.txt Mon Feb 27 13:03:19 2017 -0500
@@ -0,0 +1,4 @@
+Population FlowCL Query Nb Results Link to PDF Top Result Label Top Result Score Top Result CL
+1 CD4+CCR3-CD8-CCR7- 5+ PDF effector CD4-positive, alpha-beta T cell 0.167 CL_0001044
+2 CD4+CCR3-CD8-CCR7- 5+ PDF effector CD4-positive, alpha-beta T cell 0.167 CL_0001044
+3 CD4+CCR3-CD8+CCR7lo 5+ PDF mature CD8_alpha-negative CD11b-positive dendritic cell 2)" -0.357 CL_0001003
diff -r 000000000000 -r 8547aedf1350 profile_cl/test-data/flowcl_pop01.pdf
Binary file profile_cl/test-data/flowcl_pop01.pdf has changed
diff -r 000000000000 -r 8547aedf1350 profile_cl/test-data/flowcl_pop01.txt
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/profile_cl/test-data/flowcl_pop01.txt Mon Feb 27 13:03:19 2017 -0500
@@ -0,0 +1,44 @@
+ [,1]
+Short marker names "CD4+CCR3-CD8-CCR7-"
+Ontology marker names "CD4 molecule, C-C chemokine receptor type 3, T cell receptor"
+ "co-receptor CD8, C-C chemokine receptor type 7"
+Experiment markers "CD4,CCR3,CD8,CCR7"
+Ontology exper. names "CD4 molecule, C-C chemokine receptor type 3, T cell receptor"
+ "co-receptor CD8, C-C chemokine receptor type 7"
+Successful Match? "No"
+Marker ID "1) PR_000001004, PR_000025402, PR_000001203 2) PR_000001004,"
+ "PR_000025402, PR_000001203 3) PR_000001004, PR_000025402,"
+ "PR_000001203 4) PR_000001004, PR_000025402 5) PR_000001004,"
+ "PR_000025402 + more"
+Marker Label "1) CD4 molecule, T cell receptor co-receptor CD8, C-C chemokine"
+ "receptor type 7 2) CD4 molecule, T cell receptor co-receptor CD8, C-C"
+ "chemokine receptor type 7 3) CD4 molecule, T cell receptor"
+ "co-receptor CD8, C-C chemokine receptor type 7 4) CD4 molecule, T"
+ "cell receptor co-receptor CD8 5) CD4 molecule, T cell receptor"
+ "co-receptor CD8 + more"
+Marker Key "1) { CCR3- } CD8-, CCR7-, CD4+ ( ) [ alpha-beta T cell receptor"
+ "complex+, receptor-type tyrosine-protein phosphatase C isoform"
+ "CD45RA+, CD3+ ] 2) { CCR3- } CD8-, CCR7-, CD4+ ( ) [ alpha-beta T"
+ "cell receptor complex+, C-X-C chemokine receptor type 5+, inducible"
+ "T-cell costimulator+, programmed cell death protein 1+, CD3+ ] 3) {"
+ "CCR3- } CD8-, CCR7-, CD4+ ( ) [ interleukin-2 receptor subunit"
+ "alpha-, receptor-type tyrosine-protein phosphatase C isoform CD45RA-,"
+ "alpha-beta T cell receptor complex+, receptor-type tyrosine-protein"
+ "phosphatase C isoform CD45RO+, interleukin-7 receptor subunit alpha+,"
+ "CD3+, CD44 moleculehi, interleukin-2 receptor subunit betahi ] 4) {"
+ "CCR3-, CCR7- } CD8-, CD4+ ( ) [ L-selectin-, interleukin-7 receptor"
+ "subunit alpha-, alpha-beta T cell receptor complex+, CD69 molecule+,"
+ "interleukin-2 receptor subunit alpha+, ADP-ribosyl cyclase 1+, MHC"
+ "class II histocompatibility antigen alpha chain HLA-DRA+, CD3+ ] 5) {"
+ "CCR3-, CCR7- } CD8-, CD4+ ( ) [ interleukin-2 receptor subunit"
+ "alpha-, alpha-beta T cell receptor complex+, receptor-type"
+ "tyrosine-protein phosphatase C isoform CD45RO+, interleukin-7"
+ "receptor subunit alpha+, CD3+, CD44 moleculehi, interleukin-2"
+ "receptor subunit betahi ] + more"
+Score (Out of 1) "1) 0.167 2) 0.125 3) 0.091 4) -0.2 5) -0.222 + more"
+Cell ID "1) CL_0001044 2) CL_0002038 3) CL_0000905 4) CL_0001043 5) CL_0000897"
+ "+ more"
+Cell Label "1) effector CD4-positive, alpha-beta T cell 2) T follicular helper"
+ "cell 3) effector memory CD4-positive, alpha-beta T cell 4) activated"
+ "CD4-positive, alpha-beta T cell, human 5) CD4-positive, alpha-beta"
+ "memory T cell + more"
diff -r 000000000000 -r 8547aedf1350 profile_cl/test-data/flowcl_pop03.pdf
Binary file profile_cl/test-data/flowcl_pop03.pdf has changed
diff -r 000000000000 -r 8547aedf1350 profile_cl/test-data/flowcl_pop03.txt
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/profile_cl/test-data/flowcl_pop03.txt Mon Feb 27 13:03:19 2017 -0500
@@ -0,0 +1,43 @@
+ [,1]
+Short marker names "CD4+CCR3-CD8+CCR7lo"
+Ontology marker names "CD4 molecule, C-C chemokine receptor type 3, T cell receptor"
+ "co-receptor CD8, C-C chemokine receptor type 7"
+Experiment markers "CD4,CCR3,CD8,CCR7"
+Ontology exper. names "CD4 molecule, C-C chemokine receptor type 3, T cell receptor"
+ "co-receptor CD8, C-C chemokine receptor type 7"
+Successful Match? "No"
+Marker ID "1) PR_000001004 2) PR_000001004 3) PR_000001004 4) PR_000001004 5)"
+ "PR_000001004, PR_000025402 + more"
+Marker Label "1) CD4 molecule 2) CD4 molecule 3) CD4 molecule 4) CD4 molecule 5)"
+ "CD4 molecule, T cell receptor co-receptor CD8 + more"
+Marker Key "1) { CD8+, CCR3-, CCR7lo } CD4+ ( ) [ CD19 molecule-, CD34 molecule-,"
+ "CD34 molecule-, neural cell adhesion molecule 1-, membrane-spanning"
+ "4-domains subfamily A member 1-, lymphocyte antigen 75-, T-cell"
+ "surface glycoprotein CD8 alpha chain-, CD83 molecule+, integrin"
+ "alpha-M+, MHC class II protein complexhi, CD86 moleculehi, CD80"
+ "moleculehi, integrin alpha-Xhi ] 2) { CD8+, CCR3-, CCR7lo } CD4+ ( )"
+ "[ CD19 molecule-, CD34 molecule-, CD34 molecule-, neural cell"
+ "adhesion molecule 1-, membrane-spanning 4-domains subfamily A member"
+ "1-, lymphocyte antigen 75-, T-cell surface glycoprotein CD8 alpha"
+ "chain-, integrin alpha-M+, MHC class II protein complexlo, CD86"
+ "moleculelo, CD80 moleculelo, integrin alpha-Xhi ] 3) { CD8+, CCR3-,"
+ "CCR7lo } CD4+ ( ) [ CD19 molecule-, CD34 molecule-, CD34 molecule-,"
+ "neural cell adhesion molecule 1-, membrane-spanning 4-domains"
+ "subfamily A member 1-, integrin alpha-M-, T-cell surface glycoprotein"
+ "CD8 alpha chain-, receptor-type tyrosine-protein phosphatase C"
+ "isoform CD45R+, lymphocyte antigen 6G+, integrin alpha-Xlo ] 4) {"
+ "CD8+, CCR3-, CCR7lo } CD4+ ( ) [ CD19 molecule-, CD34 molecule-, CD34"
+ "molecule-, neural cell adhesion molecule 1-, membrane-spanning"
+ "4-domains subfamily A member 1-, integrin alpha-M-, receptor-type"
+ "tyrosine-protein phosphatase C isoform CD45R+, lymphocyte antigen"
+ "6G+, T-cell surface glycoprotein CD8 alpha chain+, integrin alpha-Xlo"
+ "] 5) { CCR3-, CCR7lo } CD4+, CD8+ ( ) [ alpha-beta T cell receptor"
+ "complexlo, CD69 molecule+ ] + more"
+Score (Out of 1) "1) -0.357 2) -0.385 3) -0.455 4) -0.455 5) -0.5 + more"
+Cell ID "1) CL_0001003 2) CL_0000997 3) CL_0002455 4) CL_0002456 5) CL_0002429"
+ "+ more"
+Cell Label "1) mature CD8_alpha-negative CD11b-positive dendritic cell 2)"
+ "immature CD8_alpha-negative CD11b-positive dendritic cell 3)"
+ "CD8_alpha-negative plasmactyoid dendritic cell 4) CD8_alpha-positive"
+ "plasmactyoid dendritic cell 5) CD69-positive double-positive"
+ "thymocyte + more"
diff -r 000000000000 -r 8547aedf1350 profile_cl/test-data/input.flowscore
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/profile_cl/test-data/input.flowscore Mon Feb 27 13:03:19 2017 -0500
@@ -0,0 +1,4 @@
+Population_ID FSC SSC CD4 CCR3 CD8 CCR7 Count Percentage
+1 1 1 3 1 1 1 1000 5
+2 1 1 3 1 1 1 1000 5
+3 1 1 3 1 3 2 1000 5
diff -r 000000000000 -r 8547aedf1350 profile_cl/test-data/out.html
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/profile_cl/test-data/out.html Mon Feb 27 13:03:19 2017 -0500
@@ -0,0 +1,91 @@
+
+
+
+FLOCK Populations Ontology
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
diff -r 000000000000 -r 8547aedf1350 profile_cl/test-data/scores.txt
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/profile_cl/test-data/scores.txt Mon Feb 27 13:03:19 2017 -0500
@@ -0,0 +1,4 @@
+Population_ID FSC SSC CD4 CCR3 CD8 CCR7 Count Percentage
+1 1 1 3 1 1 1 1000 5
+2 1 1 3 1 1 1 1000 5
+3 1 1 3 1 3 2 1000 5