Performance tests

performance/compare.py

@@ -0,0 +1,50 @@
+# This script requires pv.
+
+# https://www.ivarch.com/programs/pv.shtml
+
+# There is a Homebrew formula to install pv on macOS.
+
+# This script also depends on the simulation data:
+# make -C data
+
+import subprocess
+import sys
+
+
+def run_time_pv(command: str):
+    print(command)
+    subprocess.run(f"time {command} | pv > /dev/null", shell=True)
+    print()
+
+
+def run_bcftools(command: str, dataset_name: str):
+    run_time_pv(f"bcftools {command} data/{dataset_name}.vcf.gz")
+
+
+def run_vcztools(command: str, dataset_name: str):
+    run_time_pv(f"vcztools {command} data/{dataset_name}.vcz")
+
+
+if __name__ == "__main__":
+    commands = [
+        "view",
+        "view -s tsk_7068,tsk_8769,tsk_8820",
+        r"query -f '%CHROM %POS %REF %ALT{0}\n'",
+        r"query -f '%CHROM:%POS\n' -i 'POS=49887394 | POS=50816415'",
+        "view -s '' --force-samples",
+    ]
+    dataset = "sim_10k"
+
+    if len(sys.argv) == 2 and sys.argv[1].isnumeric():
+        index = int(sys.argv[1])
+        command = commands[index]
+        run_bcftools(command, dataset)
+        run_vcztools(command, dataset)
+    elif len(sys.argv) >= 2:
+        command = " ".join(sys.argv[1:])
+        run_bcftools(command, dataset)
+        run_vcztools(command, dataset)
+    else:
+        for command in commands:
+            run_bcftools(command, dataset)
+            run_vcztools(command, dataset)

performance/data/.gitignore

@@ -0,0 +1,3 @@
+*
+!Makefile
+!requirements.txt

performance/data/Makefile

@@ -0,0 +1,29 @@
+# The make recipes require bcftools and bgzip.
+
+# https://samtools.github.io/bcftools/howtos/install.html
+# https://www.htslib.org/doc/bgzip.html
+
+# On macOS, there are Homebrew formulas for bcftools and htslib,
+# which contains bgzip.
+
+# The Python requirements are listed in requirements.txt:
+# pip install -r requirements.txt
+
+.PHONY: all clean
+
+all: sim_10k.vcz
+
+sim_10k.ts:
+	stdpopsim HomSap -c chr22 -o sim_10k.ts pop_0:10000
+
+sim_10k.vcf.gz: sim_10k.ts
+	tskit vcf sim_10k.ts | bgzip > sim_10k.vcf.gz
+
+sim_10k.vcf.gz.csi: sim_10k.vcf.gz
+	bcftools index sim_10k.vcf.gz
+
+sim_10k.vcz: sim_10k.vcf.gz sim_10k.vcf.gz.csi
+	vcf2zarr convert sim_10k.vcf.gz sim_10k.vcz
+
+clean:
+	rm -rf sim_10k.*

performance/data/requirements.txt

@@ -0,0 +1,3 @@
+stdpopsim
+tskit
+bio2zarr