Merge branch 'latest' of github.com:dib-lab/genome-grist into latest

environment.yml

@@ -6,4 +6,4 @@ channels:
 dependencies:
  - mamba
  - screed>=1.0.5,<2
- - sourmash>=4.1,<5
+ - sourmash>=4.1.1,<5

genome_grist/conf/Snakefile

@@ -590,6 +590,7 @@ rule sourmash_prefetch_gather_wc:
         db = SOURMASH_DB_LIST,
     output:
         matches = outdir + "/genbank/{sample}.x.genbank.prefetch.zip",
+        csv = outdir + "/genbank/{sample}.x.genbank.prefetch.csv",
     resources:
         mem_mb=int(PREFETCH_MEMORY / 1e3)
     conda: "env/sourmash.yml"
@@ -601,7 +602,8 @@ rule sourmash_prefetch_gather_wc:
         echo "DB is {input.db}"
         sourmash prefetch {input.sig} {input.db} \
           --save-matches {output.matches} -k {params.ksize} \
-          --threshold-bp={params.threshold_bp} {params.moltype}
+          --threshold-bp={params.threshold_bp} {params.moltype} \
+          -o {output.csv}
     """
 
 # run sourmash search x genbank and find anything matching.

genome_grist/conf/env/sourmash.yml

@@ -5,7 +5,7 @@ channels:
 dependencies:
  - python=3.7
  - screed
- - sourmash>=4.1,<5
+ - sourmash>=4.1.1,<5
  - pip
  - pip:
    - git+https://github.com/dib-lab/genome-grist.git#egg=genome-grist

genome_grist/conf/env/sra.yml

@@ -4,5 +4,5 @@ channels:
  - bioconda
  - defaults
 dependencies:
- - sra-tools=2.11.0
+ - sra-tools=2.10.0
  - seqtk=1.3

genome_grist/gather_split.py

@@ -44,7 +44,7 @@ def main():
         notify(f"Downsampling query from scaled={query_mh.scaled} to {scaled}")
         query_mh = query_mh.downsample(scaled=int(scaled))
 
-    unknown_mh = copy.copy(query_mh)
+    unknown_mh = query_mh.to_mutable()
 
     notify(f"Loaded {len(query_mh.hashes)} hashes from '{args.query_file}'")
 

tests/test-data/HSMA33MX.conf

@@ -1,4 +1,4 @@
-sample: HSMA33MX
+sample: SRR5950647
 outdir: outputs.test.HSMA33MX
 sourmash_database_glob_pattern: tests/test-data/HSMA33MX-subset.x.genbank.matches.sig
 metagenome_trim_memory: 1e9