only run CheckIlluminaDirectory w/ LINK_LOCS=true if HiSeqX/4k and Ch…

…eckIlluminaDirectory failed previously (#641)

* remove broken symlinks in runs; call CheckIlluminaDirectory

HiSeq4000 and HiSeq X runs write out a single “s.locs” file to
Data/Intensities, rather than per-tile cluster location files within
Data/Intensities/L*/*.locs. Related to this, CheckIlluminaDirectory
with LINK_LOCS=‘true’ creates symlinks from where per-tile location
files would be to the absolute path of the s.locs file. This can cause
problems when moving a run to a new system, since the absolute paths
will be incorrect. This addition checks to see if s.locs is present. If
so, broken symlinks within the Data/Intensitites/L* are removed, and
LINK_LOCS is specified for the CheckIlluminaDirectory call that now
happens before demultiplexing. This call ensures the run looks correct,
and in the case of HiSeq4000/HiSeqX runs, also creates symlinks for
per-tile location files using absolute paths on the current system.
This is all something of a workaround for Broad-derived
HiSeq4000/HiSeqX runs, which are delivered with brittle absolute
symlinks. With this addition, runs from Broad walkup or elsewhere are
more cloud-compatible. In connection with this, a PR has been opened in
the Picard repository to request that LINK_LOCS can create relative
symlinks rather than absolute:
broadinstitute/picard#877

* remove_broken_symlinks -> find_broken_symlinks, with logging changes

remove_broken_symlinks has been renamed to find_broken_symlinks since
it now returns a list of broken links rather than removing them
directly. The function in util.file is now silent, and logging is
performed where the function is called in illumina.py.

* pass link_locs as python bool

* run CheckIlluminaDirectory w/o `LINK_LOCS=true` and then, if it fails, and if it’s a HiSeq X / 4000, run it again with `LINK_LOCS=true`

run CheckIlluminaDirectory w/o `LINK_LOCS=true` and then, if it fails,
and if it’s a HiSeq X / 4000, run it again with `LINK_LOCS=true`

illumina.py

@@ -114,26 +114,37 @@ def main_illumina_demux(args):
     # These links may break if the run directory is moved.
     # We should begin by removing broken links, if present,
     # and call CheckIlluminaDirectory ourselves if a 's.locs'
-    # file is present
-    if os.path.exists(os.path.join(illumina.get_intensities_dir(), "s.locs")):
-        # recurse to remove broken links in directory
-        log.info("This run has an 's.locs' file; checking for and removing broken per-tile symlinks...")
-        broken_links = util.file.find_broken_symlinks(illumina.get_intensities_dir())
-        if len(broken_links):
-            for lpath in broken_links:
-                log.info("Removing broken symlink: %s", lpath)
-                os.unlink(lpath)
-
-        # call CheckIlluminaDirectory with LINK_LOCS=true
-        link_locs=True
-
-    log.info("Checking run directory with Picard...")
-    tools.picard.CheckIlluminaDirectoryTool().execute(
-        illumina.get_BCLdir(),
-        args.lane,
-        illumina.get_RunInfo().get_read_structure(),
-        link_locs=link_locs
-    )
+    # file is present, but only if the directory check fails
+    # since link_locs=true tries to create symlinks even if they 
+    # (or the files) already exist
+    try:
+        tools.picard.CheckIlluminaDirectoryTool().execute(
+            illumina.get_BCLdir(),
+            args.lane,
+            illumina.get_RunInfo().get_read_structure(),
+            link_locs=link_locs
+        )
+    except subprocess.CalledProcessError as e:
+        log.warning("CheckIlluminaDirectory failed for %s", illumina.get_BCLdir())
+        if os.path.exists(os.path.join(illumina.get_intensities_dir(), "s.locs")):
+            # recurse to remove broken links in directory
+            log.info("This run has an 's.locs' file; checking for and removing broken per-tile symlinks...")
+            broken_links = util.file.find_broken_symlinks(illumina.get_intensities_dir())
+            if len(broken_links):
+                for lpath in broken_links:
+                    log.info("Removing broken symlink: %s", lpath)
+                    os.unlink(lpath)
+
+            # call CheckIlluminaDirectory with LINK_LOCS=true
+            link_locs=True
+
+            log.info("Checking run directory with Picard...")
+            tools.picard.CheckIlluminaDirectoryTool().execute(
+                illumina.get_BCLdir(),
+                args.lane,
+                illumina.get_RunInfo().get_read_structure(),
+                link_locs=link_locs
+            )
 
     # Picard ExtractIlluminaBarcodes
     extract_input = util.file.mkstempfname('.txt', prefix='.'.join(['barcodeData', flowcell, str(args.lane)]))