diff --git a/config/datatypes_conf.xml.sample b/config/datatypes_conf.xml.sample
index 7c696c543de..7bbbb1b0a1b 100644
--- a/config/datatypes_conf.xml.sample
+++ b/config/datatypes_conf.xml.sample
@@ -155,6 +155,11 @@
       <display file="rviewer/bed.xml" inherit="true"/>
       <display file="igv/interval_as_bed.xml" inherit="true"/>
     </datatype>
+
+    <!-- ISA data types -->
+    <datatype extension="isa-tab" type="galaxy.datatypes.isa:IsaTab" mimetype="application/isa-tools" display_in_upload="true" description="ISA-Tab data type." description_url="https://isa-tools.org"/>
+    <datatype extension="isa-json" type="galaxy.datatypes.isa:IsaJson" mimetype="application/isa-tools" display_in_upload="true" description="ISA-JSON data type." description_url="https://isa-tools.org"/>
+
     <datatype extension="picard_interval_list" type="galaxy.datatypes.tabular:Tabular" subclass="true" display_in_upload="true">
       <converter file="picard_interval_list_to_bed6_converter.xml" target_datatype="bed6"/>
     </datatype>
@@ -837,6 +842,7 @@
     <sniffer type="galaxy.datatypes.images:Xpm"/>
     <sniffer type="galaxy.datatypes.images:Eps"/>
     <sniffer type="galaxy.datatypes.images:Rast"/>
+
     <!--
     Keep this commented until the sniff method in the assembly.py
     module is fixed to not read the entire file.
diff --git a/lib/galaxy/datatypes/isa.py b/lib/galaxy/datatypes/isa.py
new file mode 100644
index 00000000000..7b1cfa6f44a
--- /dev/null
+++ b/lib/galaxy/datatypes/isa.py
@@ -0,0 +1,383 @@
+"""
+ISA datatype
+
+See https://github.com/ISA-tools
+
+"""
+
+from __future__ import print_function
+
+import json
+import logging
+import os
+import os.path
+import re
+import shutil
+import sys
+import tempfile
+from cgi import escape
+from json import dumps  # noqa: F401
+
+# Imports isatab after turning off warnings inside logger settings to avoid pandas warning making uploads fail.
+logging.getLogger("isatools.isatab").setLevel(logging.ERROR)
+from isatools import isajson
+from isatools import isatab_meta
+
+from galaxy import model
+from galaxy import util
+from galaxy.datatypes import data
+from galaxy.util.compression_utils import CompressedFile
+from galaxy.util.sanitize_html import sanitize_html
+
+# CONSTANTS {{{1
+################################################################
+
+# Main files regex
+JSON_FILE_REGEX = re.compile(r"^.*\.json$", flags=re.IGNORECASE)
+INVESTIGATION_FILE_REGEX = re.compile(r"^i_\w+\.txt$", flags=re.IGNORECASE)
+
+# The name of the ISA archive (compressed file) as saved inside Galaxy
+ISA_ARCHIVE_NAME = "archive"
+
+# Set max number of lines of the history peek
+_MAX_LINES_HISTORY_PEEK = 11
+
+# Configure logger {{{1
+################################################################
+
+logger = logging.getLogger(__name__)
+
+# Function for opening correctly a CSV file for csv.reader() for both Python 2 and 3 {{{1
+################################################################
+
+
+def utf8_text_file_open(path):
+    if sys.version_info[0] < 3:
+        fp = open(path, 'rb')
+    else:
+        fp = open(path, 'r', newline='', encoding='utf8')
+    return fp
+
+
+# ISA class {{{1
+################################################################
+
+class _Isa(data.Data):
+    """ Base class for implementing ISA datatypes """
+    composite_type = 'auto_primary_file'
+    allow_datatype_change = False
+    is_binary = True
+    _main_file_regex = None
+
+    # Make investigation instance {{{2
+    ################################################################
+
+    def _make_investigation_instance(self, filename):
+        raise NotImplementedError()
+
+    # Constructor {{{2
+    ################################################################
+
+    def __init__(self, main_file_regex, **kwd):
+        super(_Isa, self).__init__(**kwd)
+        self._main_file_regex = main_file_regex
+
+        # Add the archive file as the only composite file
+        self.add_composite_file(ISA_ARCHIVE_NAME, is_binary=True, optional=True)
+
+    # Get ISA folder path {{{2
+    ################################################################
+
+    def _get_isa_folder_path(self, dataset):
+
+        isa_folder = None
+
+        if dataset:
+            if isinstance(dataset, model.Dataset):
+                isa_folder = dataset.extra_files_path
+            if isinstance(dataset, model.HistoryDatasetAssociation):
+                # XXX With this loop the dataset name is reset inside the history to the ISA archive ID. Why?
+                for attr, value in dataset.__dict__.iteritems():
+                    if str(attr) == '_metadata_collection':
+                        datatype = value.parent.datatype  # noqa: F841
+                isa_folder = dataset.dataset.extra_files_path
+
+        if isa_folder is None:
+            raise Exception('Unvalid dataset object, or no extra files path found for this dataset.')
+
+        return isa_folder
+
+    # Get main file {{{2
+    ################################################################
+
+    def _get_main_file(self, dataset):
+        """Get the main file of the ISA archive. Either the investigation file i_*.txt for ISA-Tab, or the JSON file for ISA-JSON."""
+
+        main_file = None
+        isa_folder = self._get_isa_folder_path(dataset)
+
+        if os.path.exists(isa_folder):
+
+            # Get ISA archive older
+            isa_files = os.listdir(isa_folder)
+
+            # Try to find main file
+            main_file = self._find_main_file_in_archive(isa_files)
+
+            if main_file is None:
+                raise Exception('Invalid ISA archive. No main file found.')
+
+            # Make full path
+            main_file = os.path.join(isa_folder, main_file)
+
+        return main_file
+
+    # Get investigation {{{2
+    ################################################################
+
+    def _get_investigation(self, dataset):
+        """Create a contained instance specific to the exact ISA type (Tab or Json).
+           We will use it to parse and access information from the archive."""
+
+        investigation = None
+        main_file = self._get_main_file(dataset)
+        if main_file is not None:
+            investigation = self._make_investigation_instance(main_file)
+
+        return investigation
+
+    # Find main file in archive {{{2
+    ################################################################
+
+    def _find_main_file_in_archive(self, files_list):
+        """Find the main file inside the ISA archive."""
+
+        found_file = None
+
+        for f in files_list:
+            match = self._main_file_regex.match(f)
+            if match:
+                if found_file is None:
+                    found_file = match.group()
+                else:
+                    raise Exception('More than one file match the pattern "', str(self._main_file_regex), '" to identify the investigation file')
+
+        return found_file
+
+    # Set peek {{{2
+    ################################################################
+
+    def set_peek(self, dataset, is_multi_byte=False):
+        """Set the peek and blurb text. Get first lines of the main file and set it as the peek."""
+
+        main_file = self._get_main_file(dataset)
+
+        if main_file is None:
+            raise RuntimeError("Unable to find the main file within the 'files_path' folder")
+
+        # Read first lines of main file
+        with open(main_file, "r") as f:
+            data = []
+            for line in f:
+                if len(data) < _MAX_LINES_HISTORY_PEEK:
+                    data.append(line)
+                else:
+                    break
+            if not dataset.dataset.purged and data:
+                dataset.peek = json.dumps({"data": data})
+                dataset.blurb = 'data'
+            else:
+                dataset.peek = 'file does not exist'
+                dataset.blurb = 'file purged from disk'
+
+    # Display peek {{{2
+    ################################################################
+
+    def display_peek(self, dataset):
+        """Create the HTML table used for displaying peek, from the peek text found by set_peek() method."""
+
+        out = ['<table cellspacing="0" cellpadding="3">']
+        try:
+            if not dataset.peek:
+                dataset.set_peek()
+            json_data = json.loads(dataset.peek)
+            for line in json_data["data"]:
+                line = line.strip()
+                if not line:
+                    continue
+                out.append('<tr><td>%s</td></tr>' % escape(util.unicodify(line, 'utf-8')))
+            out.append('</table>')
+            out = "".join(out)
+        except Exception as exc:
+            out = "Can't create peek %s" % str(exc)
+        return out
+
+    # Generate primary file {{{2
+    ################################################################
+
+    def generate_primary_file(self, dataset=None):
+        """Generate the primary file. It is an HTML file containing description of the composite dataset
+           as well as a list of the composite files that it contains."""
+
+        if dataset:
+            rval = ['<html><head><title>ISA Dataset </title></head><p/>']
+            if hasattr(dataset, "extra_files_path"):
+                rval.append('<div>ISA Dataset composed of the following files:<p/><ul>')
+                for cmp_file in os.listdir(dataset.extra_files_path):
+                    rval.append('<li><a href="%s" type="text/plain">%s</a></li>' % (cmp_file, escape(cmp_file)))
+                rval.append('</ul></div></html>')
+            else:
+                rval.append('<div>ISA Dataset is empty!<p/><ul>')
+            return "\n".join(rval)
+        return "<div>No dataset available</div>"
+
+    # Dataset content needs grooming {{{2
+    ################################################################
+
+    def dataset_content_needs_grooming(self, file_name):
+        """This function is called on an output dataset file after the content is initially generated."""
+        return os.path.basename(file_name) == ISA_ARCHIVE_NAME
+
+    # Groom dataset content {{{2
+    ################################################################
+
+    def groom_dataset_content(self, file_name):
+        """This method is called by Galaxy to extract files contained in a composite data type."""
+        # XXX Is the right place to extract files? Should this step not be a cleaning step instead?
+        # Could extracting be done earlier and composite files declared as files contained inside the archive
+        # instead of the archive itself?
+
+        # extract basename and folder of the current file whose content has to be groomed
+        basename = os.path.basename(file_name)
+        output_path = os.path.dirname(file_name)
+        # extract archive if the file corresponds to the ISA archive
+        if basename == ISA_ARCHIVE_NAME:
+            # perform extraction
+            # For some ZIP files CompressedFile::extract() extract the file inside <output_folder>/<file_name> instead of outputing it inside <output_folder>. So we first create a temporary folder, extract inside it, and move content to final destination.
+            temp_folder = tempfile.mkdtemp()
+            CompressedFile(file_name).extract(temp_folder)
+            shutil.rmtree(output_path)
+            extracted_files = os.listdir(temp_folder)
+            logger.debug(' '.join(extracted_files))
+            if len(extracted_files) == 0:
+                os.makedirs(output_path)
+                shutil.rmtree(temp_folder)
+            elif len(extracted_files) == 1 and os.path.isdir(os.path.join(temp_folder, extracted_files[0])):
+                shutil.move(os.path.join(temp_folder, extracted_files[0]), output_path)
+                shutil.rmtree(temp_folder)
+            else:
+                shutil.move(temp_folder, output_path)
+
+    # Display data {{{2
+    ################################################################
+
+    def display_data(self, trans, dataset, preview=False, filename=None, to_ext=None, offset=None, ck_size=None, **kwd):
+        """Downloads the ISA dataset if `preview` is `False`;
+           if `preview` is `True`, it returns a preview of the ISA dataset as a HTML page.
+           The preview is triggered when user clicks on the eye icon of the composite dataset."""
+
+        # if it is not required a preview use the default behaviour of `display_data`
+        if not preview:
+            return super(_Isa, self).display_data(trans, dataset, preview, filename, to_ext, **kwd)
+
+        # prepare the preview of the ISA dataset
+        investigation = self._get_investigation(dataset)
+        if investigation is None:
+            html = """<html><header><title>Error while reading ISA archive.</title></header>
+                   <body>
+                        <h1>An error occured while reading content of ISA archive.</h1>
+                        <p>If you have tried to load your archive with the uploader by selecting isa-tab as composite data type, then try to load it again with isa-json instead. Conversely, if you have tried to load your archive with the uploader by selecting isa-json as composite data type, then try isa-tab instead.</p>
+                        <p>You may also try to look into your zip file in order to find out if this is a proper ISA archive. If you see a file i_Investigation.txt inside, then it is an ISA-Tab archive. If you see a file with extension .json inside, then it is an ISA-JSON archive. If you see nothing like that, then either your ISA archive is corrupted, or it is not an ISA archive.</p>
+                   </body></html>"""
+        else:
+            html = '<html><body>'
+            html += '<h1>{0} {1}</h1>'.format(investigation.title, investigation.identifier)
+
+            # Loop on all studies
+            for study in investigation.studies:
+                html += '<h2>Study %s</h2>' % study.identifier
+                html += '<h3>%s</h3>' % study.title
+                html += '<p>%s</p>' % study.description
+                html += '<p>Submitted the %s</p>' % study.submission_date
+                html += '<p>Released on %s</p>' % study.public_release_date
+
+                html += '<p>Experimental factors used: %s</p>' % ', '.join([x.name for x in study.factors])
+
+                # Loop on all assays of this study
+                for assay in study.assays:
+                    html += '<h3>Assay %s</h3>' % assay.filename
+                    html += '<p>Measurement type: %s</p>' % assay.measurement_type.term  # OntologyAnnotation
+                    html += '<p>Technology type: %s</p>' % assay.technology_type.term    # OntologyAnnotation
+                    html += '<p>Technology platform: %s</p>' % assay.technology_platform
+                    if assay.data_files is not None:
+                        html += '<p>Data files:</p>'
+                        html += '<ul>'
+                        for data_file in assay.data_files:
+                            if data_file.filename != '':
+                                html += '<li>' + escape(util.unicodify(str(data_file.filename), 'utf-8')) + ' - ' + escape(util.unicodify(str(data_file.label), 'utf-8')) + '</li>'
+                        html += '</ul>'
+
+            html += '</body></html>'
+
+        # Set mime type
+        mime = 'text/html'
+        self._clean_and_set_mime_type(trans, mime)
+
+        return sanitize_html(html).encode('utf-8')
+
+
+# ISA-Tab class {{{1
+################################################################
+
+class IsaTab(_Isa):
+    file_ext = "isa-tab"
+
+    # Constructor {{{2
+    ################################################################
+
+    def __init__(self, **kwd):
+        super(IsaTab, self).__init__(main_file_regex=INVESTIGATION_FILE_REGEX, **kwd)
+
+    # Make investigation instance {{{2
+    ################################################################
+
+    def _make_investigation_instance(self, filename):
+
+        # Parse ISA-Tab investigation file
+        parser = isatab_meta.InvestigationParser()
+        isa_dir = os.path.dirname(filename)
+        fp = utf8_text_file_open(filename)
+        parser.parse(fp)
+        for study in parser.isa.studies:
+            s_parser = isatab_meta.LazyStudySampleTableParser(parser.isa)
+            s_parser.parse(os.path.join(isa_dir, study.filename))
+            for assay in study.assays:
+                a_parser = isatab_meta.LazyAssayTableParser(parser.isa)
+                a_parser.parse(os.path.join(isa_dir, assay.filename))
+        isa = parser.isa
+
+        return isa
+
+
+# ISA-JSON class {{{1
+################################################################
+
+class IsaJson(_Isa):
+    file_ext = "isa-json"
+
+    # Constructor {{{2
+    ################################################################
+
+    def __init__(self, **kwd):
+        super(IsaJson, self).__init__(main_file_regex=JSON_FILE_REGEX, **kwd)
+
+    # Make investigation instance {{{2
+    ################################################################
+
+    def _make_investigation_instance(self, filename):
+
+        # Parse JSON file
+        fp = utf8_text_file_open(filename)
+        isa = isajson.load(fp)
+
+        return isa
diff --git a/lib/galaxy/dependencies/pipfiles/default/pinned-requirements.txt b/lib/galaxy/dependencies/pipfiles/default/pinned-requirements.txt
index da7429017a9..e9006ab405f 100644
--- a/lib/galaxy/dependencies/pipfiles/default/pinned-requirements.txt
+++ b/lib/galaxy/dependencies/pipfiles/default/pinned-requirements.txt
@@ -60,6 +60,7 @@ h5py==2.8.0
 html5lib==1.0.1
 idna==2.7
 ipaddress==1.0.22; python_version < '3.3'
+isa-rwval==0.10.4
 iso8601==0.1.12
 isodate==0.6.0
 jmespath==0.9.3
diff --git a/test-data/MTBLS6.zip b/test-data/MTBLS6.zip
new file mode 100644
index 00000000000..61543c14d47
Binary files /dev/null and b/test-data/MTBLS6.zip differ
diff --git a/test/api/test_tools_upload.py b/test/api/test_tools_upload.py
index b11cda9c6cd..c8296c4df80 100644
--- a/test/api/test_tools_upload.py
+++ b/test/api/test_tools_upload.py
@@ -240,6 +240,14 @@ class ToolsUploadTestCase(api.ApiTestCase):
             roadmaps_content = self._get_roadmaps_content(history_id, dataset)
             assert roadmaps_content.strip() == "roadmaps\ncontent", roadmaps_content
 
+    @skip_without_datatype("isa-tab")
+    def test_composite_datatype_isatab(self):
+        isatab_zip_path = TestDataResolver().get_filename("MTBLS6.zip")
+        details = self._upload_and_get_details(open(isatab_zip_path, "rb"), file_type="isa-tab")
+        assert details["state"] == "ok"
+        assert details["file_ext"] == "isa-tab", details
+        assert details["file_size"] == 85, details
+
     def test_upload_dbkey(self):
         with self.dataset_populator.test_history() as history_id:
             payload = self.dataset_populator.upload_payload(history_id, "Test123", dbkey="hg19")
diff --git a/tools/data_source/upload.py b/tools/data_source/upload.py
index 5c047e70ff8..920025180b0 100644
--- a/tools/data_source/upload.py
+++ b/tools/data_source/upload.py
@@ -171,7 +171,15 @@ def add_file(dataset, registry, output_path):
     return info
 
 
-def add_composite_file(dataset, output_path, files_path):
+def add_composite_file(dataset, registry, output_path, files_path):
+
+    # Find data type
+    if dataset.file_type is not None:
+        try:
+            datatype = registry.get_datatype_by_extension(dataset.file_type)
+        except Exception as e:
+            print("Unable to instantiate the datatype object for the file type '%s'" % dataset.file_type)
+
     if dataset.composite_files:
         os.mkdir(files_path)
         for name, value in dataset.composite_files.items():
@@ -195,9 +203,17 @@ def add_composite_file(dataset, output_path, files_path):
                         sniff.convert_newlines_sep2tabs(dp, tmp_dir=tmpdir, tmp_prefix=tmp_prefix)
                     else:
                         sniff.convert_newlines(dp, tmp_dir=tmpdir, tmp_prefix=tmp_prefix)
-                shutil.move(dp, os.path.join(files_path, name))
+
+                # move the file to its final destination
+                file_output_path = os.path.join(files_path, name)
+                shutil.move(dp, file_output_path)
+                # groom the dataset file content if required by the corresponding datatype definition
+                if datatype.dataset_content_needs_grooming(file_output_path):
+                    datatype.groom_dataset_content(file_output_path)
+
     # Move the dataset to its "real" path
     shutil.move(dataset.primary_file, output_path)
+
     # Write the job info
     return dict(type='dataset',
                 dataset_id=dataset.dataset_id,
@@ -263,7 +279,7 @@ def __main__():
         try:
             if dataset.type == 'composite':
                 files_path = output_paths[int(dataset.dataset_id)][1]
-                metadata.append(add_composite_file(dataset, output_path, files_path))
+                metadata.append(add_composite_file(dataset, registry, output_path, files_path))
             else:
                 metadata.append(add_file(dataset, registry, output_path))
         except UploadProblemException as e: