From 5fdfece97fb2d50a10eab5004a6467ec0097ece8 Mon Sep 17 00:00:00 2001
From: Pjotr Prins
Date: Fri, 6 Nov 2020 11:19:28 +0000
Subject: Uploader script improvements

---
 bh20sequploader/main.py     | 5 +++--
 bh20sequploader/qc_fasta.py | 9 +++++----
 doc/INSTALL.md              | 8 +++++++-
 3 files changed, 15 insertions(+), 7 deletions(-)

diff --git a/bh20sequploader/main.py b/bh20sequploader/main.py
index f89b458..ea0fa70 100644
--- a/bh20sequploader/main.py
+++ b/bh20sequploader/main.py
@@ -49,7 +49,7 @@ sequence for enough overlap with the reference genome
                 failed = True
     except Exception as e:
         log.exception("Failed metadata QC")
-        failed = True
+        failed = True # continue with the FASTA checker
 
     target = []
     try:
@@ -64,13 +64,14 @@ sequence for enough overlap with the reference genome
             target[1] = ("reads_2."+target[1][0][6:], target[1][1], target[1][2])
 
         if do_qc and target[0][2] == 'text/fasta' and sample_id != target[0][1]:
-            raise ValueError("The sample_id field in the metadata must be the same as the FASTA header")
+            raise ValueError(f"The sample_id field in the metadata ({sample_id}) must be the same as the FASTA header ({target[0][1]})")
 
     except Exception as e:
         log.exception("Failed sequence QC")
         failed = True
 
     if failed:
+        log.debug("Bailing out!")
         exit(1)
 
     return target
diff --git a/bh20sequploader/qc_fasta.py b/bh20sequploader/qc_fasta.py
index f567f0a..814fb3e 100644
--- a/bh20sequploader/qc_fasta.py
+++ b/bh20sequploader/qc_fasta.py
@@ -66,7 +66,8 @@ def qc_fasta(arg_sequence, check_with_mimimap2=True):
 
                     similarity = 0
                     try:
-                        cmd = ["minimap2", "-c -x asm20", tmp1.name, tmp2.name]
+                        log.debug("Trying to run minimap2")
+                        cmd = ["minimap2", "-c", "-x", "asm20", tmp1.name, tmp2.name]
                         logging.info("QC checking similarity to reference")
                         logging.info(" ".join(cmd))
                         result = subprocess.run(cmd, stdout=subprocess.PIPE)
@@ -83,9 +84,7 @@ def qc_fasta(arg_sequence, check_with_mimimap2=True):
 
                     if similarity < 70.0:
                         raise ValueError(
-                            "QC fail for {}: alignment to reference was less than 70%% (was %2.2f%%)".format(
-                                seqlabel, similarity
-                            ))
+                            f"QC fail for {seqlabel}: alignment to reference was less than 70% (was {similarity})")
 
         return "sequence.fasta" + gz, seqlabel, seq_type
     elif seq_type == "text/fastq":
@@ -93,4 +92,6 @@ def qc_fasta(arg_sequence, check_with_mimimap2=True):
         sequence.detach()
         return "reads.fastq" + gz, seqlabel, seq_type
     else:
+        log.debug(seqlabel)
+        log.debug(seq_type)
         raise ValueError("Sequence file ({}) does not look like a DNA FASTA or FASTQ".format(arg_sequence))
diff --git a/doc/INSTALL.md b/doc/INSTALL.md
index f54c8f2..45aca0f 100644
--- a/doc/INSTALL.md
+++ b/doc/INSTALL.md
@@ -31,7 +31,7 @@ arvados-python-client-2.0.1 ciso8601-2.1.3 future-0.18.2 google-api-python-clien
 3. Run the tool directly with
 
 ```sh
-guix environment guix --ad-hoc git python openssl python-pycurl python-magic nss-certs python-pyshex -- python3 bh20sequploader/main.py example/sequence.fasta example/maximum_metadata_example.yaml
+guix environment guix --ad-hoc git python openssl python-pycurl python-magic nss-certs python-pyshex -- python3 bh20sequploader/main.py example/maximum_metadata_example.yaml example/sequence.fasta
 ```
 
 Note that python-pyshex is packaged in
@@ -44,6 +44,12 @@ repository. E.g.
 env GUIX_PACKAGE_PATH=~/iwrk/opensource/guix/guix-bioinformatics/ ~/opt/guix/bin/guix environment -C guix --ad-hoc git python python-flask python-pyyaml python-pycurl python-magic  nss-certs python-pyshex python-pyyaml --network openssl python-pyshex python-pyshexc minimap2 python-schema-salad python-arvados-python-client --share=/export/tmp -- env TMPDIR=/export/tmp python3 bh20sequploader/main.py --help
 ```
 
+Latest successful Guix run
+
+```sh
+env GUIX_PACKAGE_PATH=~/iwrk/opensource/guix/guix-bioinformatics/ ~/opt/guix/bin/guix environment guix --ad-hoc git python openssl python-pycurl python-magic nss-certs python-pyshex python-arvados-python-client python-schema-salad minimap2 -- python3 bh20sequploader/main.py  scripts/uthsc_samples/yaml/AL_UT14.yaml scripts/uthsc_samples/yaml/AL_UT14.fa
+```
+
 ### Using the Web Uploader
 
 To run the web uploader in a GNU Guix environment/container run it with something like
-- 
cgit v1.2.3