New environment for additional benchmarks.

additional_benchmarks/README.md

@@ -0,0 +1,7 @@
+# Additional Benchmarks
+
+## Environment
+
+ mamba create -n granges_benchmark -c conda-forge -c bioconda --file requirements.txt --yes
+ pip install snakemake-executor-plugin-slurm
+

additional_benchmarks/Snakefile

@@ -1,8 +1,11 @@
 import numpy as np
 
+RSCRIPT = "~/.conda/envs/granges_benchmark/bin/R"
 GRANGES = "../target/release/granges"
 SEQLENS = "../tests_data/hg38_seqlens.tsv"
-NREPS = 1
+NREPS = 50
+
+localrules: granges
 
 rule granges:
  output: GRANGES
@@ -14,6 +17,10 @@ rule granges:
 rule random_bed:
  input: seqlens=SEQLENS, granges=GRANGES
  output: "random_bed/{size}__{rep}.bed.gz"
+ resources:
+ runtime=30,
+ mem_mb_per_cpu=1800,
+ cpus_per_task=28
  shell:
  """
  {input.granges} random-bed --sort --num {wildcards.size} {input.seqlens} | gzip > {output}
@@ -24,6 +31,10 @@ rule granges_filter:
  genome=SEQLENS, granges=GRANGES
  output: "results/granges_filter__size_{size}.bed"
  benchmark: repeat("benchmarks/granges_filter__size_{size}.tsv", NREPS)
+ resources:
+ runtime=30,
+ mem_mb_per_cpu=1800,
+ cpus_per_task=28
  shell:
  """
  {input.granges} filter --left {input.a} --right {input.b} --genome {input.genome} > {output}
@@ -34,6 +45,10 @@ rule bedtools_intersect:
  input: a="random_bed/{size}__A.bed.gz", b="random_bed/{size}__B.bed.gz"
  output: "results/bedtools_intersect__size_{size}.bed"
  benchmark: repeat("benchmarks/bedtools_intersect__size_{size}.tsv", NREPS)
+ resources:
+ runtime=30,
+ mem_mb_per_cpu=1800,
+ cpus_per_task=28
  shell:
  """
  bedtools intersect -a {input.a} -b {input.b} > {output}
@@ -42,14 +57,20 @@ rule bedtools_intersect:
 rule plyranges_join_overlap_inner:
  input: a="random_bed/{size}__A.bed.gz", b="random_bed/{size}__B.bed.gz"
  output: "results/plyranges_join_overlap_inner__size_{size}.bed"
+ params: rscript = RSCRIPT
  benchmark: repeat("benchmarks/plyranges_join_overlap_inner__size_{size}.tsv", NREPS)
+ resources:
+ runtime=300,
+ mem_mb_per_cpu=1800,
+ cpus_per_task=28
  shell:
  """
- Rscript Rscripts/plyranges_join_overlap_inner.r {input.a} {input.b} > {output}
+ {params.rscript} Rscripts/plyranges_join_overlap_inner.r {input.a} {input.b} > {output}
  """
 
-tools = ["bedtools_intersect", "granges_filter", "plyranges_join_overlap_inner"]
-sizes = np.logspace(3, 9, 10)
+#tools = ["bedtools_intersect", "granges_filter", "plyranges_join_overlap_inner"]
+tools = ["bedtools_intersect", "granges_filter"]
+sizes = np.logspace(3, 9, 10).astype('int')
 
 all_benches = expand("results/{tool}__size_{size}.bed", tool=tools, size=sizes)
 

additional_benchmarks/requirements.txt

@@ -0,0 +1,5 @@
+snakemake
+numpy
+scipy
+matplotlib
+R