Demo/benchmark mashing-pumpkins to build minhash sketches

demo_benchmark_mashingp.sh

#!/bin/sh

echo
echo '---------------'
echo ' test w/ FASTA ' 
echo '---------------'

testchrom=chr1.fa.gz

if [ ! -f "${testchrom}" ]; then
   wget ftp://hgdownload.cse.ucsc.edu/goldenPath/hg38/chromosomes/${testchrom};
fi
   
testfile=${testchrom}
   
echo '- mashing-pumpkins'
time python -m mashingpumpkins.demo.cmdline --format=FASTA --ncpu=3 --maxsize=500 ${testfile}
mv ${testfile}.sig.json ${testfile}_mashingpumpkins.sig.json

echo '---'
echo '- sourmash'
time sourmash compute --force --dna -o ${testfile}_sourmash.sig.json ${testfile}

echo '***'
echo 'The same signature, just faster ?'
sourmash search ${testfile}_mashingpumpkins.sig.json ${testfile}_sourmash.sig.json



echo
echo '---------------'
echo ' test w/ FASTQ ' 
echo '---------------'


testfastq=DRR013190.fastq.gz

ksize=31
maxsize=1000

if [ ! -f "${testfastq}" ]; then
   wget ftp://ftp.sra.ebi.ac.uk/vol1/fastq/DRR013/DRR013190/${testfastq}
fi
   
testfile=${testfastq}

echo '- mashing-pumpkins'

for ncpu in 2 3 4
do
    echo '  Using ' ${ncpu} ' cores.'
    time python -m mashingpumpkins.demo.cmdline --ncpu=${ncpu} --format=FASTQ --ksize=${ksize} --maxsize=${maxsize} ${testfile};
    echo '  --'
done
mv ${testfile}.sig.json ${testfile}_mashingpumpkins.sig.json

echo '---'
echo '- sourmash'
time sourmash compute --force --dna -n ${maxsize} --k ${ksize} -o ${testfile}_sourmash.sig.json ${testfile}

echo '***'
echo 'The same signature, just faster ?'
sourmash search --k ${ksize} -n ${maxsize} --dna --no-protein ${testfile}_mashingpumpkins.sig.json ${testfile}_sourmash.sig.json


echo '  Using faster parser and ' ${ncpu} ' cores.'
time python -m mashingpumpkins.demo.cmdline --ncpu=3 --parser=fastqandfurious --format=FASTQ --ksize=${ksize} --maxsize=${maxsize} ${testfile};
echo '  --'


echo
echo '----------------------'
echo ' test w/ larger FASTQ ' 
echo '----------------------'


testfastq=DRR065801.fastq.gz

if [ ! -f "${testfastq}" ]; then
   wget ftp://ftp.sra.ebi.ac.uk/vol1/fastq/DRR065/DRR065801/${testfastq}
fi

testfile=${testfastq}
echo '- mashing-pumpkins'
time python -m mashingpumpkins.demo.cmdline --ncpu=3 --format=FASTQ --ksize=${ksize} --maxsize=${maxsize} ${testfile};

echo '  Using faster parser'
time python -m mashingpumpkins.demo.cmdline --ncpu=3 --parser=fastqandfurious --format=FASTQ --ksize=${ksize} --maxsize=${maxsize} ${testfile};
echo '  --'


echo '---'
echo '- sourmash'
time sourmash compute --force --dna -n ${maxsize} --k ${ksize} -o ${testfile}_sourmash.sig.json ${testfile}