Merge branch 'telatin-tests'

tests/10_benchmark.sh

@@ -5,6 +5,13 @@ set -e
 thisDir=$(dirname $0);
 export PATH=$thisDir/../target/release:$PATH
 
+# Check whether hyperfine is installed
+if ! command -v hyperfine &> /dev/null
+then
+    echo "hyperfine could not be found. It is required for benchmarking."
+    exit
+fi
+
 # Hyperfine parameters
 # Locally, just run one time per test but in the cloud, boost it to ten
 num_runs=10

tests/benchmark_sample.sh

@@ -10,7 +10,8 @@ export PATH=$thisDir/../target/release:$PATH
 hyperfine --export-json=$reportsDir/sample.json --warmup 2 --shell $SHELL --runs $num_runs \
   -n "Fasten sample" "cat $large_R1 | fasten_sample --frequency 0.1" \
   -n "seqkit sample" "cat $large_R1 | seqkit sample --proportion 0.1" \
-  -n "Seqtk sample" "seqtk seq -f 0.1 $large_R1"; 
+  -n "Seqtk sample" "seqtk seq -f 0.1 $large_R1" \
+  -n "Seqfu sample" "seqfu cat --skip 10 $large_R1"; 
 
-plot_whisker.py --title "subsample reads (reps=$num_runs)" --labels "fasten sample,seqkit sample,seqtk sample" --output $reportsDir/sample.json.png $reportsDir/sample.json
+plot_whisker.py --title "subsample reads (reps=$num_runs)" --labels "fasten sample,seqkit sample,seqtk sample,seqfu cat" --output $reportsDir/sample.json.png $reportsDir/sample.json
 

tests/fasten_combine.sh

@@ -16,15 +16,15 @@ if [ "$reads_not_collapsed" != "$original_reads" ]; then
 fi
 
 reads_collapsed=$(cat $INPUT $INPUT $INPUT | ./target/debug/fasten_combine | ./target/debug/fasten_metrics --each-read)
-total_quals=$(echo "$reads_collapsed" | cut -f 3 | tail -n +2 | paste -sd+ | bc -l)
+total_quals=$(echo "$reads_collapsed" | cut -f 3 | tail -n +2 | awk '{if(NR>1) printf "+"; printf $1} END{print "\n";}' | bc -l)
 if [ "$total_quals" != "259.31" ]; then
   echo "Test failed for total expected quality when collapsing three sets of reads"
   exit 1
 fi
 
 pe_collapsed=$(cat $INPUT $INPUT $INPUT | ./target/debug/fasten_combine --paired-end | ./target/debug/fasten_metrics --each-read)
-IDs=$(echo "$pe_collapsed" | cut -f 1 | tail -n +2 | paste -sd+)
-pe_quals=$(echo "$pe_collapsed" | cut -f 3 | tail -n +2 | paste -sd+ | bc -l)
+IDs=$(echo "$pe_collapsed" | cut -f 1 | tail -n +2 |  awk '{if(NR>1) printf "+"; printf $1}')
+pe_quals=$(echo "$pe_collapsed" | cut -f 3 | tail -n +2 |  awk '{if(NR>1) printf "+"; printf $1} END{print "\n";}' | bc -l)
 if [ "$IDs" != "1/1+1/2+2/1+2/2+3/1+3/2+4/1+4/2" ]; then
   echo "Test failed for total expected quality when collapsing three sets of reads using --paired-end"
   exit 1

tests/fasten_convert.sh

@@ -0,0 +1,23 @@
+#!/usr/bin/env bash
+# Minimal test suite for fasten  (telatin 2024)
+
+THIS_SCRIPT_DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )"
+THIS_SCRIPT_NAME=$(basename "$0" | sed 's/\.sh//g')
+source "${THIS_SCRIPT_DIR}/test_functions.sh"
+
+IN_FILE="${THIS_SCRIPT_DIR}/../testdata/four_reads.fastq"
+
+"$BIN" --out-format FASTA < "$IN_FILE" > "$TEST_TMP_FILE"
+FASTA_COUNT=$(grep -c ">" "$TEST_TMP_FILE")
+FASTQ_COUNT=$(grep -c "^@" "$TEST_TMP_FILE")
+equal "$FASTA_COUNT" "4" "Testing that the output is in FASTA format"
+equal "$FASTQ_COUNT" "0" "Testing that the output is in not FASTQ format"
+
+"$BIN" --out-format FASTQ --in-format FASTA < "$TEST_TMP_FILE" > "$TEST_TMP_FILE.2"
+
+FASTA_COUNT=$(grep -c "^>" "$TEST_TMP_FILE.2")
+FASTQ_COUNT=$(grep -c "^@r" "$TEST_TMP_FILE.2")
+
+equal "$FASTQ_COUNT" "4" "Testing that the output is in not FASTA format"
+equal "$FASTQ_COUNT" "4" "Testing that the output is in FASTQ format"
+rm "$TEST_TMP_FILE.2"

tests/fasten_inspect.sh

@@ -0,0 +1,6 @@
+#!/usr/bin/env bash
+# Minimal test suite for fasten  (telatin 2024)
+
+THIS_SCRIPT_DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )"
+THIS_SCRIPT_NAME=$(basename "$0" | sed 's/\.sh//g')
+source "${THIS_SCRIPT_DIR}/test_functions.sh"

tests/fasten_mutate.sh

@@ -0,0 +1,18 @@
+#!/usr/bin/env bash
+# Minimal test suite for fasten mutate (telatin 2024)
+
+THIS_SCRIPT_DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )"
+THIS_SCRIPT_NAME=$(basename "$0" | sed 's/\.sh//g')
+source "${THIS_SCRIPT_DIR}/test_functions.sh"
+
+INFILE="${THIS_SCRIPT_DIR}/../testdata/four_reads.fastq"
+
+
+"$BIN" --snps 1 < "$INFILE" > "$TEST_TMP_FILE"
+
+MD5_IN_EXPECTED="8a08ae75226dfacd60f6fe2a1000f100"
+MD5=$(getmd5 "$TEST_TMP_FILE" | cut -f 1 -d " ")
+MD5_IN=$(getmd5 "$INFILE" | cut -f 1 -d " ")
+
+equal "$MD5_IN" "$MD5_IN_EXPECTED" "Testing that the input file wasnt changed"
+different "$MD5" "$MD5_IN" "Testing that the output is different from the input"

tests/fasten_normalize.sh

@@ -0,0 +1,6 @@
+#!/usr/bin/env bash
+# Minimal test suite for fasten normalise (telatin 2024)
+
+THIS_SCRIPT_DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )"
+THIS_SCRIPT_NAME=$(basename "$0" | sed 's/\.sh//g')
+source "${THIS_SCRIPT_DIR}/test_functions.sh"

tests/fasten_progress.sh

@@ -0,0 +1,39 @@
+#!/usr/bin/env bash
+# Minimal test suite for fasten progress (telatin 2024)
+
+THIS_SCRIPT_DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )"
+THIS_SCRIPT_NAME=$(basename "$0" | sed 's/\.sh//g')
+source "${THIS_SCRIPT_DIR}/test_functions.sh"
+
+GOT_PERL=$(which perl)
+
+if [ -z "$GOT_PERL" ]; then
+    echo "Perl not found, skipping test"
+    exit 0
+fi
+
+
+## Here we test that STDOUT is passed using --print
+
+# shellcheck disable=SC2016
+TOT=$("$GOT_PERL" -e 'my $c=0;for (1..1000) { 
+    $c++;
+    print "\@fasten_test$c\nAAA\n+\nIII\n";
+    sleep 0.1;
+ }' | "$BIN" --id "test-suite" --print | grep -c 'fasten_test' | grep -w 1000) 
+
+equal "$TOT" "1000" "Testing sort order of 1000 reads"
+
+
+## Here we test the final message 
+# shellcheck disable=SC2016
+"$GOT_PERL" -e 'my $c=0;for (1..1000) { 
+    $c++;
+    print "\@fasten_test$c\nAAA\n+\nIII\n";
+    sleep 0.1;
+ }' | "$BIN" --id "test-suite" 2> "$TEST_TMP_FILE"
+
+END=$(grep "Finished" "$TEST_TMP_FILE" | cut -f 3 -d ":")
+equal "$END" " Finished progress on 4000 reads" "Testing progress output"
+
+done_testing

tests/fasten_regex.sh

@@ -26,12 +26,12 @@ if [ "$(wc -l <<< "$pe_filtered")" -ne 24 ]; then
   exit 1
 fi
 
-if [ "$(echo "$pe_filtered" | ./target/debug/fasten_metrics --each-read | tail -n +2 | cut -f 1 | paste -sd+)" != "read0/1+read0/2+read1/1+read1/2+read2/1+read2/2" ]; then
+if [ "$(echo "$pe_filtered" | ./target/debug/fasten_metrics --each-read | tail -n +2 | cut -f 1 |  awk '{if(NR>1) printf "+"; printf $1}' )" != "read0/1+read0/2+read1/1+read1/2+read2/1+read2/2" ]; then
   echo "ERROR filtering for the right read names"
   exit 1
 fi
 
-if [ "$(echo "$r1_filtered" | ./target/debug/fasten_regex --regex read1 --which ID | ./target/debug/fasten_metrics --each-read | tail -n +2 | cut -f 1 | paste -sd+)" != "read1/1+read1/2" ]; then
+if [ "$(echo "$r1_filtered" | ./target/debug/fasten_regex --regex read1 --which ID | ./target/debug/fasten_metrics --each-read | tail -n +2 | cut -f 1 |  awk '{if(NR>1) printf "+"; printf $1}' )" != "read1/1+read1/2" ]; then
   echo "ERROR running regex on IDs for read1"
   exit 1
 fi

tests/fasten_sort.sh

@@ -0,0 +1,12 @@
+#!/usr/bin/env bash
+# Minimal test suite for fasten sort (telatin 2024)
+
+THIS_SCRIPT_DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )"
+THIS_SCRIPT_NAME=$(basename "$0" | sed 's/\.sh//g')
+source "${THIS_SCRIPT_DIR}/test_functions.sh"
+
+# Test fasten_sort with two reads
+FIRST=$(echo -e "@ciao\nAAA\n+\nIII\n@andrea\nCCC\n+\nEEE" | $BIN | head -n 1)
+equal "$FIRST" "@andrea" "Testing sort order of two reads"
+
+done_testing

tests/fasten_trim.sh

@@ -10,7 +10,7 @@ if [ "$reads_not_trimmed" != "$original_reads" ]; then
   exit 1
 fi
 
-onebase=$(./target/debug/fasten_trim --first-base 3 --last-base 4 < testdata/four_reads.pe.fastq | perl -lane 'print if($i++ % 4 == 1);' | paste -sd'_')
+onebase=$(./target/debug/fasten_trim --first-base 3 --last-base 4 < testdata/four_reads.pe.fastq | perl -lane 'print if($i++ % 4 == 1);' |awk 'NR > 1 { printf "_"; } { printf $1; } END { printf "\n"; }')
 shouldbe="T_T_G_A_C_A_C_A"
 if [ "$onebase" != "$shouldbe" ]; then
   echo "ERROR trimming to the third base"

tests/lib/benchmark.sh

@@ -51,12 +51,17 @@ zcat $large_interleaved | fasten_sort --sort-by GC --paired-end | gzip -c > $lar
 
 which bbnorm.sh
 which fasten_clean
+which seqfu
 
 # Version information
 seqtk 2>&1 | grep -i version | sed 's/^/seqtk /'
 seqkit version | grep -m 1 v
 fasten_clean --version
 fastq_to_fasta -h | grep "Part of FASTX"
 bbnorm.sh version 2>&1 | grep 'BBMap version'
+seqfu --version
 
+# hyperfine
+which hyperfine
+which plot_whisker.py || echo "WARNING: plot_whisker.py from hyperfine not found in path: will not be able to plot graphs."
 

tests/test_functions.sh

@@ -0,0 +1,91 @@
+#!/usr/bin/env bash
+
+# This should be sourced by other test scripts: die if not
+if [ -z "$THIS_SCRIPT_DIR" ]; then
+    echo "ERROR: test_functions.sh should be sourced by other test scripts"
+    exit 1
+fi
+TEST_TMP_FILE=$(mktemp)
+NUM=0
+FAIL=0
+RED='\033[0;31m'
+GREEN='\033[0;32m'
+NC='\033[0m' # No Color
+BIN=$(readlink -f "${THIS_SCRIPT_DIR}/../target/release/${THIS_SCRIPT_NAME}")
+DEB_BIN=$(readlink -f "${THIS_SCRIPT_DIR}/../target/debug/${THIS_SCRIPT_NAME}")
+
+
+echo -e " *** ${GREEN}Testing $THIS_SCRIPT_NAME${NC} (test: $TEST_TMP_FILE)*** "
+
+
+
+function test {
+    NUM=$((NUM+1))
+    local msg="$1"
+    local condition=$2
+    if [ "$condition" ]; then
+        echo -e "${GREEN}OK${NC}\t$NUM: $msg"
+    else
+        FAIL=$((FAIL+1))
+        echo -e "${RED}FAIL${NC}\t$NUM: $msg"
+    fi
+}
+
+function equal {
+    local got="$1"
+    local expected="$2"
+    local msg="$3"
+    NUM=$((NUM+1))
+    if [ "$got" == "$expected" ]; then
+        echo -e "${GREEN}OK${NC}\t$NUM: $msg [$got]"
+    else
+        FAIL=$((FAIL+1))
+        echo -e "${RED}FAIL${NC}\t$NUM: $msg"
+        echo -e "\tGot: $got"
+        echo -e "\tExpected: $expected"
+    fi
+}
+
+function different {
+    local got="$1"
+    local expected="$2"
+    local msg="$3"
+    NUM=$((NUM+1))
+    if [ "$got" != "$expected" ]; then
+        echo -e "${GREEN}OK${NC}\t$NUM: $msg [$got != $expected]"
+    else
+        FAIL=$((FAIL+1))
+        echo -e "${RED}FAIL${NC}\t$NUM: $msg"
+        echo -e "\tGot: $got"
+        echo -e "\tequals to: $expected"
+    fi
+}
+
+function getmd5 {
+    # use md5sum on Linux, md5 on OSX
+    if [ "$(uname)" == "Darwin" ]; then
+        md5 -q "$1"
+    else
+        md5sum "$1" | cut -f 1 -d " "
+    fi
+}
+function done_testing {
+    if [ -e "$TEST_TMP_FILE" ]; then
+        rm "$TEST_TMP_FILE"
+    fi
+    if [ "$FAIL" -eq 0 ]; then
+        echo -e "${GREEN}OK!${NC}\tAll $NUM tests passed${NC}"
+        exit 0
+    else
+        echo -e "${RED}$FAIL/$NUM errors${NC}\ttests failed${NC}"
+        exit 1
+    fi
+}
+
+test "Release binary $BIN" "-e $BIN"
+test "Release debug $DEB_BIN" "-e $DEB_BIN"
+test "Release binary --help" "$DEB_BIN --help"
+test "Debug binary --help" "$DEB_BIN --help"
+
+test "Release binary --version" "$DEB_BIN --version"
+test "Debug binary --version" "$DEB_BIN --version"