Add download of fasta of query contigs assembled as chromosomes, Fixes #36

src/dgenies/lib/functions.py

@@ -17,6 +17,8 @@ ALLOWED_EXTENSIONS = ['fa', 'fasta', 'fna', 'fa.gz', 'fasta.gz', 'fna.gz']
 
 class Functions:
 
+    config = AppConfigReader()
+
     @staticmethod
     def allowed_file(filename):
         return '.' in filename and \
@@ -140,14 +142,13 @@ class Functions:
             return j1.email
 
     @staticmethod
-    def send_fasta_ready(mailer, job_name, sample_name, compressed=False):
-        config = AppConfigReader()
-        web_url = config.web_url
+    def send_fasta_ready(mailer, job_name, sample_name, compressed=False, path="fasta-query", status="success"):
+        web_url = Functions.config.web_url
         with open(os.path.join(os.path.dirname(os.path.realpath(__file__)), "mail_templates", "dl_fasta.html")) \
                 as t_file:
             template = Template(t_file.read())
-            message_html = template.render(job_name=job_name, status="success", url_base=web_url,
-                                           sample_name=sample_name, compressed=compressed)
+            message_html = template.render(job_name=job_name, status=status, url_base=web_url,
+                                           sample_name=sample_name, compressed=compressed, path=path)
         message = "D-Genies\n\n" \
                   "Job %s - Download fasta\n\n" % job_name
         message += "Query fasta file for job %s (query: %s) is ready to download.\n" % (job_name, sample_name)
@@ -156,6 +157,8 @@ class Functions:
         mailer.send_mail([Functions.get_mail_for_job(job_name)], "Job %s - Download fasta" % job_name, message,
                          message_html)
 
+
+
     @staticmethod
     def sort_fasta(job_name, fasta_file, index_file, lock_file, compress=False, mailer=None):
         index, sample_name = Functions.read_index(index_file)
@@ -229,4 +232,3 @@ class Functions:
                 "time_elapsed": Functions.get_readable_time(item.job.time_elapsed)
             })
         return items
-

src/dgenies/lib/mail_templates/dl_fasta.html

@@ -29,7 +29,7 @@
 <p>Hi,</p>
 {% if status == "success" %}
     <p>Query fasta file for job {{ job_name }} (query: {{ sample_name }}) is ready to download.<br/>
-        You can <a href="{{ url_base }}/fasta-query/{{ job_name }}/{{ sample_name }}.fasta{% if compressed %}.gz{% endif %}">click here</a> to download it.</p>
+        You can <a href="{{ url_base }}/{{ path }}/{{ job_name }}/{{ sample_name }}.fasta{% if compressed %}.gz{% endif %}">click here</a> to download it.</p>
 {% else %}
     <p>Build of query fasta file for job {{ job_name }} has failed. You can try again. If the problem persists, please contact the support.</p>
 {% endif %}

src/dgenies/lib/paf.py

@@ -5,24 +5,29 @@ import shutil
 from math import sqrt
 from numpy import mean
 from pathlib import Path
-import matplotlib as mpl
-mpl.use('Agg')
-from matplotlib import pyplot as plt
 import json
 from dgenies.bin.index import Index
+from dgenies.lib.functions import Functions
 from intervaltree import IntervalTree
+import matplotlib as mpl
+mpl.use('Agg')
+from matplotlib import pyplot as plt
+from Bio import SeqIO
+from Bio.Seq import Seq
+from Bio.SeqRecord import SeqRecord
 
 
 class Paf:
     limit_idy = [0.25, 0.5, 0.75]
     max_nb_lines = 100000
 
-    def __init__(self, paf: str, idx_q: str, idx_t: str, auto_parse: bool=True):
+    def __init__(self, paf: str, idx_q: str, idx_t: str, auto_parse: bool=True, mailer=None, id_job=None):
         self.paf = paf
         self.idx_q = idx_q
         self.idx_t = idx_t
         self.sorted = False
-        if os.path.exists(os.path.join(os.path.dirname(paf), ".sorted")):
+        self.data_dir = os.path.dirname(paf)
+        if os.path.exists(os.path.join(self.data_dir, ".sorted")):
             self.paf += ".sorted"
             self.idx_q += ".sorted"
             self.sorted = True
@@ -41,6 +46,8 @@ class Paf:
         self.name_t = None
         self.parsed = False
         self.error = False
+        self.mailer = mailer
+        self.id_job = id_job
 
         if auto_parse:
             self.parse_paf()
@@ -444,6 +451,27 @@ class Paf:
                 query_on_target[contig] = None
         return query_on_target
 
+    def get_queries_on_target_association(self):
+        """
+        For each target, get the list of queries associated to it
+        :return:
+        """
+        gravity_contig = self.compute_gravity_contigs()[0]
+        queries_on_target = {}
+        for contig, chr_blocks in gravity_contig.items():
+            # Find best block:
+            max_number = 0
+            max_chr = None
+            for chrm, size in chr_blocks.items():
+                if size > max_number:
+                    max_number = size
+                    max_chr = chrm
+            if max_chr is not None:
+                if max_chr not in queries_on_target:
+                    queries_on_target[max_chr] = []
+                queries_on_target[max_chr].append(contig)
+        return queries_on_target
+
     def build_query_on_target_association_file(self):
         """
         For each query, get the best matching chromosome and save it to a CSV file.
@@ -645,3 +673,54 @@ class Paf:
                 txt = summary_file.read()
                 return json.loads(txt)
         return None
+
+    def build_query_chr_as_reference(self):
+        try:
+            if not self.sorted:
+                raise Exception("Contigs must be sorted to do that!")
+            contigs_assoc = self.get_queries_on_target_association()
+            with open(os.path.join(self.data_dir, ".query")) as query_file:
+                query_fasta = query_file.read().strip("\n")
+            if not os.path.isfile(query_fasta):
+                raise Exception("Query fasta does not exists")
+            uncompressed = False
+            if query_fasta.endswith(".gz"):
+                uncompressed = True
+                query_fasta = Functions.uncompress(query_fasta)
+            query_f = SeqIO.index(query_fasta, "fasta")
+            o_fasta = os.path.join(os.path.dirname(query_fasta), "as_reference_" + os.path.basename(query_fasta))
+            mapped_queries = set()
+            with open(o_fasta, "w") as out:
+                for target in self.t_order:
+                    if target in contigs_assoc:
+                        queries = sorted(contigs_assoc[target], key=lambda x: self.q_order.index(x))
+                        seq = SeqRecord(Seq(""))
+                        for query in queries:
+                            mapped_queries.add(query)
+                            new_seq = query_f[query]
+                            if self.q_reversed[query]:
+                                new_seq = new_seq.reverse_complement()
+                            seq += new_seq
+                            seq += 100 * "N"
+                        seq = seq[:-100]
+                        seq.id = seq.name = seq.description = target
+                        SeqIO.write(seq, out, "fasta")
+                for contig in self.q_order:
+                    if contig not in mapped_queries:
+                        seq = query_f[contig]
+                        seq.id += "_unaligned"
+                        SeqIO.write(seq, out, "fasta")
+            if uncompressed:
+                os.remove(query_fasta)
+            status = "success"
+        except Exception:
+            o_fasta = None
+            status="fail"
+
+        Functions.send_fasta_ready(mailer=self.mailer,
+                                   job_name=self.id_job,
+                                   sample_name="as_reference_" + os.path.basename(query_fasta).rsplit(".")[0],
+                                   compressed=False,
+                                   path="download",
+                                   status=status)
+        return o_fasta

src/dgenies/static/js/d3.boxplot.js

@@ -103,10 +103,12 @@ d3.boxplot.launch = function(res, update=false, noise_change=false) {
     if (res["sorted"]) {
         $("input#sort-contigs").val("Undo sort");
         $("#export").find("select option[value=4]").show();
+        $("#export").find("select option[value=8]").show();
     }
     else {
         $("input#sort-contigs").val("Sort contigs");
         $("#export").find("select option[value=4]").hide();
+        $("#export").find("select option[value=8]").hide();
     }
     d3.boxplot.name_x = res["name_x"];
     d3.boxplot.name_y = res["name_y"];

src/dgenies/static/js/dgenies.result.export.js

@@ -141,6 +141,19 @@ dgenies.result.export.export_no_association_file = function (to) {
         })
 };
 
+dgenies.result.export.export_query_as_reference_fasta = function() {
+    dgenies.post(`/build-query-as-reference/${dgenies.result.id_res}`,
+        {},
+        function (data, success) {
+            if (data["success"]) {
+                dgenies.notify("You will receive a mail soon with the link to download your Fasta file", "success")
+            }
+            else {
+                dgenies.notify(`An error has occurred. Please contact the support`, "fatal")
+            }
+        });
+}
+
 dgenies.result.export.export = function () {
     let select = $("form#export select");
     let selection = parseInt(select.val());
@@ -166,6 +179,9 @@ dgenies.result.export.export = function () {
             else if (selection === 7) {
                 dgenies.result.export.export_no_association_file("target");
             }
+            else if (selection === 8) {
+                dgenies.result.export.export_query_as_reference_fasta();
+            }
             else
                 dgenies.notify("Not supported yet!", "danger", 2000);
             if (!async)

src/dgenies/templates/result.html

@@ -42,6 +42,7 @@
                 <option value="2">Png</option>
                 <option value="3">Paf file</option>
                 <option value="4">Query Fasta</option>
+                <option value="8">Query assembled as reference</option>
                 <option value="5">Association table</option>
                 <option value="6">No match queries</option>
                 <option value="7">No match targets</option>

src/dgenies/views.py

@@ -331,6 +331,26 @@ def build_fasta(id_res):
                         "message": "Unable to get fasta file for query. Please contact us to report the bug"})
 
 
+@app.route('/build-query-as-reference/<id_res>', methods=['POST'])
+def get_query_as_reference(id_res):
+    paf_file = os.path.join(APP_DATA, id_res, "map.paf")
+    idx1 = os.path.join(APP_DATA, id_res, "query.idx")
+    idx2 = os.path.join(APP_DATA, id_res, "target.idx")
+    paf = Paf(paf_file, idx1, idx2, False, mailer=mailer, id_job=id_res)
+    paf.parse_paf(False, True)
+    thread = threading.Timer(0, paf.build_query_chr_as_reference)
+    thread.start()
+    return jsonify({"success": True})
+
+
+@app.route('/download/<id_res>/<filename>')
+def download_file(id_res, filename):
+    file_dl = os.path.join(APP_DATA, id_res, filename)
+    if os.path.isfile(file_dl):
+        return send_file(file_dl)
+    return abort(404)
+
+
 @app.route('/fasta-query/<id_res>', defaults={'filename': ""}, methods=['GET'])
 @app.route('/fasta-query/<id_res>/<filename>', methods=['GET'])  # Use fake URL in mail to set download file name
 def dl_fasta(id_res, filename):