hanslovsky/convert-cremi-mutex-watersheds-all.sh

## convert-cremi-mutex-watersheds-all.sh
_convert() {
    ITERATION=$1
    SETUP=$2
    THRESHOLD=$3
    SAMPLE=$4
    CONTAINER="/nrs/saalfeld/hanslovskyp/experiments/quasi-isotropic-predictions/affinities-glia/neuron_ids-unlabeled-unmask-background/predictions/CREMI/sample_${SAMPLE}.n5"
    BASE_PATH="volumes/predictions/neuron_ids-unlabeled-unmask-background/${SETUP}/${ITERATION}"
    if [ -z "${THRESHOLD}" ]; then
        SOURCE="${BASE_PATH}/mutex-watershed-merged"
    else
        SOURCE="${BASE_PATH}/mutex-watershed-threshold=${THRESHOLD}-merged"
    fi
    TARGET="${SOURCE}-multiscale"

    if [ -d "${CONTAINER}/${TARGET}" ]; then
        echo "${CONTAINER}/${TARGET} already exists. Skipping..."
        exit 0
    fi

    paintera-convert \
        to-paintera \
        --block-size=64,64,64 \
        --scale 2 2 2 2 2 \
        --output-container="${CONTAINER}" \
        --container="${CONTAINER}" \
        --dataset="${SOURCE}" \
        --overwrite-existing \
        --target-dataset="${TARGET}"
    echo "Successfully converted into dataset ${TARGET} in container ${CONTAINER}."

}

mkdir -p $HOME/.local/tmp/mutex-watersheds-eval/logs

for ITERATION in 500000; do
    for SETUP in 0 3; do
        for THRESHOLD in 0.5 '' 0.3 0.7 0.1 0.9; do
            for SAMPLE in 2 1 0 A C B; do
                _convert $ITERATION $SETUP "$THRESHOLD" $SAMPLE
            done
        done
    done
done

## eval-mutex-watersheds-all.sh
_evaluate() {
    ITERATION=$1
    SETUP=$2
    SAMPLE=$3
    THRESHOLD=$4
    if [ -z "$5" ]; then
        WITH_GLIA=''
        WITH_GLIA_ARG='--with-glia=false'
    else
        WITH_GLIA='-with-glia'
        WITH_GLIA_ARG='--with-glia=true'
    fi
    if [ -z "$THRESHOLD" ]; then
        eval-mutex-watersheds \
            --sample=$SAMPLE \
            --setup=$SETUP \
            --iteration=$ITERATION \
            --border-threshold=175 \
            --overwrite-existing \
            "${WITH_GLIA_ARG}" \
            1> "$HOME/.local/tmp/mutex-watersheds-eval/logs/eval.log.sample=${SAMPLE}_setup=${SETUP}_iteration=${ITERATION}${WITH_GLIA}" \
            2>&1
        eval-mutex-watersheds \
            --sample=$SAMPLE \
            --setup=$SETUP \
            --iteration=$ITERATION \
            --border-threshold=175 \
            --ignore-training-mask \
            --overwrite-existing \
            "${WITH_GLIA_ARG}" \
            1> "$HOME/.local/tmp/mutex-watersheds-eval/logs/eval.log.sample=${SAMPLE}_setup=${SETUP}_iteration=${ITERATION}${WITH_GLIA}_ignore-training-mask" \
            2>&1
    else
        eval-mutex-watersheds \
            --sample=$SAMPLE \
            --setup=$SETUP \
            --iteration=$ITERATION \
            --threshold=$THRESHOLD \
            --border-threshold=175 \
            --overwrite-existing \
            "${WITH_GLIA_ARG}" \
            1> "$HOME/.local/tmp/mutex-watersheds-eval/logs/eval.log.sample=${SAMPLE}_setup=${SETUP}_iteration=${ITERATION}_threshold=${THRESHOLD}${WITH_GLIA}" \
            2>&1
        eval-mutex-watersheds \
            --sample=$SAMPLE \
            --setup=$SETUP \
            --iteration=$ITERATION \
            --threshold=$THRESHOLD \
            --border-threshold=175 \
            --ignore-training-mask \
            --overwrite-existing \
            "${WITH_GLIA_ARG}" \
            1> "$HOME/.local/tmp/mutex-watersheds-eval/logs/eval.log.sample=${SAMPLE}_setup=${SETUP}_iteration=${ITERATION}_threshold=${THRESHOLD}${WITH_GLIA}_ignore-training-mask" \
            2>&1
    fi
}

_evaluate_all_thresholds() {
    ITERATION=$1
    SETUP=$2
    SAMPLE=$3
    WITH_GLIA="$4"
    for THRESHOLD in 0.5 '' 0.3 0.7 0.1 0.9; do
        _evaluate $ITERATION $SETUP $SAMPLE "$THRESHOLD" "$WITH_GLIA"
    done

}

mkdir -p $HOME/.local/tmp/mutex-watersheds-eval/logs

for ITERATION in 500000; do
    for SETUP in 0 3; do
        for SAMPLE in A B C 0 1 2; do
            _evaluate_all_thresholds $ITERATION $SETUP $SAMPLE 1 &
        done
    done
done

wait

## interesting-neurons.md

      
    Raw
  

              interesting-neurons.md
            
          
mutex-watershed-threshold=0.5-merged-multiscale

20585282
16428617
16647665
20525818
3737 potentially easy to fix merge
11168539
3788453 probably many merges
14199231
17548588
20515317
15526596
15501213 potentially easy to fix merge


## lauritzen-mutex-watershed-to-mipmap.sh
N_NODES=${N_NODES:-10}
N_EXECUTORS_PER_NODE=${N_EXECUTORS_PER_NODE:-15}

BLOCK_SIZE=${BLOCK_SIZE:-64,64,64}

# N5 group that holds integer type labels
# N5_GROUP="/nrs/saalfeld/hanslovskyp/n5-examples"
BLOCK=${BLOCK:-03}
CONTAINER="${CONTAINER:-/nrs/saalfeld/hanslovskyp/experiments/quasi-isotropic-predictions/affinities-glia/neuron_ids-unlabeled-unmask-background/predictions/lauritzen/${BLOCK}/workspace.n5}"
if [ -z "${THRESHOLD}" ]; then
    SOURCE_DATASET="${SOURCE_DATASET:-volumes/predictions/neuron_ids-unlabeled-unmask-background/0/500000/mutex-watershed-merged}"
    TARGET_DATASET="${TARGET_DATASET:-${SOURCE_DATASET}-multiscale}"
else
    SOURCE_DATASET="${SOURCE_DATASET:-volumes/predictions/neuron_ids-unlabeled-unmask-background/0/500000/mutex-watershed-threshold=${THRESHOLD}-merged}"
    TARGET_DATASET="${TARGET_DATASET:-${SOURCE_DATASET}-multiscale}"
fi
# input dataset
DATASET="filtered/volumes/labels/multiset-mipmap-with-label-lists-per-block"

JAR="${JAR:-$HOME/paintera-conversion-helper-0.9.2-shaded.jar}"
CLASS="org.janelia.saalfeldlab.paintera.conversion.PainteraConvert"

N_EXECUTORS_PER_NODE=$N_EXECUTORS_PER_NODE \
                    $HOME/flintstone/flintstone.sh ${N_NODES} $JAR $CLASS \
                    to-paintera \
                    --block-size=$BLOCK_SIZE \
                    --scale 2 2 2 2 2 2 2 2 \
                    -m -1 -1 -1
5 4 3 2 \
                    --output-container="${CONTAINER}" \
                    --container="${CONTAINER}" \
                    --dataset="${SOURCE_DATASET}" \
                    --target-dataset="${TARGET_DATASET}"

## paintera-cremi-predictions.sh
BASE_PATH="/nrs/saalfeld/hanslovskyp/experiments/quasi-isotropic-predictions/affinities-glia/neuron_ids-unlabeled-unmask-background/predictions/CREMI"
SAMPLE="$1"
PREDICTION_CONTAINER="${BASE_PATH}/sample_${SAMPLE}.n5"
SETUP="$2"
THRESHOLD="$3"
DATASET_BASE="volumes/predictions/neuron_ids-unlabeled-unmask-background/$SETUP/500000"
if [ -z "$THRESHOLD" ]; then
    MERGED_STRING="mutex-watershed-merged-multiscale"
    WITH_GLIA_STRING="mutex-watershed-merged-with-glia"
else
    MERGED_STRING="mutex-watershed-threshold=$THRESHOLD-merged-multiscale"
    WITH_GLIA_STRING="mutex-watershed-threshold=$THRESHOLD-merged-with-glia"
fi
paintera \
    --add-n5-container "/groups/saalfeld/home/hanslovskyp/data/from-arlo/interpolated-combined/sample_${SAMPLE}.n5" --resolution=36,36,360 -d volumes/raw \
    --add-n5-container "/groups/saalfeld/home/hanslovskyp/data/from-arlo/interpolated-combined/sample_${SAMPLE}.n5" --revert-array-attributes -d volumes/labels/glia-downsampled volumes/labels/neuron_ids-downsampled \
    --add-n5-container "${PREDICTION_CONTAINER}" -d "${DATASET_BASE}/affinities" "${DATASET_BASE}/glia" "${DATASET_BASE}/${MERGED_STRING}" "${DATASET_BASE}/${WITH_GLIA_STRING}" --channels 8,4,0 9,5,1 10,6,2 11,7,3

## qify-synapse-predictions.kts
#!/usr/bin/env kscript

@file:KotlinOpts("-J-Xmx100g")

@file:MavenRepository("scijava.public", "https://maven.scijava.org/content/groups/public")

@file:DependsOn("org.janelia.saalfeldlab:n5:2.1.2")
@file:DependsOn("org.janelia.saalfeldlab:n5-imglib2:3.4.1")
@file:DependsOn("org.janelia.saalfeldlab:k5:0.1.0")
@file:DependsOn("org.apache.commons:commons-compress:1.18")
@file:DependsOn("info.picocli:picocli:4.0.2")
@file:DependsOn("net.imglib2:imglib2-algorithm:0.11.1")
@file:DependsOn("net.imglib2:imglib2-realtransform:2.2.1")
// use logback to set log level programmatically
@file:DependsOn("ch.qos.logback:logback-classic:1.3.0-alpha4")
@file:DependsOn("sc.fiji:bigdataviewer-core:7.0.0")
// @file:DependsOn("org.slf4j:slf4j-api:2.0.0-alpha0")
// @file:DependsOn("org.slf4j:slf4j-simple:2.0.0-alpha0")


import java.io.File
import java.lang.Runtime
import java.lang.Thread
import java.util.concurrent.Callable
import java.util.concurrent.Executors

import kotlin.system.exitProcess

import bdv.export.Downsample
import com.google.gson.JsonElement
import net.imglib2.FinalInterval
import net.imglib2.RandomAccessibleInterval
import net.imglib2.algorithm.gauss3.Gauss3
import net.imglib2.algorithm.util.Grids
import net.imglib2.converter.Converter
import net.imglib2.converter.Converters
import net.imglib2.converter.RealDoubleConverter
import net.imglib2.img.array.ArrayImgs
import net.imglib2.interpolation.randomaccess.NLinearInterpolatorFactory
import net.imglib2.realtransform.AffineTransform3D
import net.imglib2.realtransform.RealViews
import net.imglib2.type.NativeType
import net.imglib2.type.numeric.RealType
import net.imglib2.type.numeric.integer.ByteType
import net.imglib2.type.numeric.integer.UnsignedByteType
import net.imglib2.type.numeric.real.DoubleType
import net.imglib2.type.numeric.real.FloatType
import net.imglib2.util.Fraction
import net.imglib2.util.Intervals
import net.imglib2.view.Views
import org.janelia.saalfeldlab.k5.K5
import org.janelia.saalfeldlab.n5.DatasetAttributes
import org.janelia.saalfeldlab.n5.DataType
import org.janelia.saalfeldlab.n5.GzipCompression
import org.janelia.saalfeldlab.n5.N5FSReader
import org.janelia.saalfeldlab.n5.N5FSWriter
import org.janelia.saalfeldlab.n5.N5Reader
import org.janelia.saalfeldlab.n5.imglib2.N5Utils
import picocli.CommandLine

// This is necessary to get auto-detection of n5 compression to work, for some reason.
Thread
    .currentThread()
    .setContextClassLoader(java.lang.invoke.MethodHandles.lookup().lookupClass().getClassLoader())

@CommandLine.Command(name = "qi-fy")
class CliArgs : Callable<Int> {

        @CommandLine.Option(names = ["--container", "-c"], required = true)
        lateinit var containerPath: File

        @CommandLine.Option(names = ["--dataset", "-d"], required = true)
        lateinit var dataset: String

        @CommandLine.Option(names = ["--block-size"], required = false, defaultValue = "64", split = ",")
        lateinit var blockSize: IntArray

        @CommandLine.Option(names = ["--resolution"], required = false, defaultValue = "108,108,120", split = ",")
        lateinit var resolution: DoubleArray

        @CommandLine.Option(names = ["--offset"], required = false, defaultValue = "0,0,0", split = ",")
        lateinit var offset: DoubleArray

        @CommandLine.Option(names = ["--parallelism", "-p"], required = false)
        var parallelism: Int? = null

        override fun call(): Int {
                val blockSize = if (this.blockSize.size == 3) blockSize else IntArray(3) { this.blockSize[0] }
                val sourceContainer = with (K5) { containerPath.absolutePath.n5fsWriter() }
                val sourceDataset = sourceContainer.firstMipmapLevelDataset(dataset)
                val targetDataset = "$dataset-qified"
                val scaleFactorsTargetToSource = arrayOf(Fraction(3, 1), Fraction(3, 1), Fraction(1, 3))
                val offsetsInOriginalCoordinates = arrayOf(Fraction(1, 1), Fraction(1, 1), Fraction(0, 1))
                val translationTargetToSource = offsetsInOriginalCoordinates.map { it.ratio }.toDoubleArray()
                val scaleTargetToSource = scaleFactorsTargetToSource.map { it.ratio }.toDoubleArray()
                val scaleFactorsSourceToTarget = scaleFactorsTargetToSource.map { it.clone().also { it.invert() } }
                val source = sourceContainer.datasetAsDoubleType(sourceDataset)
                println(LongArray(3) { source.dimension(it) }.map { it })
                val targetSize = LongArray(3) { (source.dimension(it) * scaleFactorsSourceToTarget[it].ratio).toLong() }
                targetSize[2] -= 2L
                println(targetSize.map { it })
                val targetInterval = FinalInterval(*targetSize)
                val extendedSource = Views.extendBorder(source)
                val interpolatedSource = Views.interpolate(extendedSource, NLinearInterpolatorFactory<DoubleType>())
                val relevantTransform = AffineTransform3D()
                // need to shift input by 1 in x and y and scale by a factor of 3 along z
                relevantTransform.set(
                    1.0, 0.0, 0.0, 1.0,
                    0.0, 1.0, 0.0, 1.0,
                    0.0, 0.0, 3.0, 0.0)
                val transformedSource = RealViews.affine(interpolatedSource, relevantTransform)

                val attributes = DatasetAttributes(
                        targetSize,
                        blockSize,
                        DataType.FLOAT64,
                        GzipCompression())
                sourceContainer.createDataset(targetDataset, attributes)
                sourceContainer.setAttribute(targetDataset, "resolution", resolution)
                sourceContainer.setAttribute(targetDataset, "offset", offset)
                val parallelism = this.parallelism ?: Runtime.getRuntime().availableProcessors()
                val es = Executors.newFixedThreadPool(parallelism)

                val factors2D = IntArray(2) { scaleFactorsTargetToSource[it].ratio.toInt() }

                val intervals = Grids.collectAllContainedIntervalsWithGridPositions(targetSize, blockSize)
                val futures = intervals.map { block ->
                        val task = Callable<Unit> {
                                val target = ArrayImgs.doubles(*Intervals.dimensionsAsLongArray(block.a))
                                val targetOffset = Views.translate(target, *Intervals.minAsLongArray(block.a))
                                println("doing stuff for $targetOffset")
                                for (index in targetOffset.min(2) .. targetOffset.max(2)) {
                                        Downsample.downsample(
                                                Views.hyperSlice(transformedSource, 2, index),
                                                Views.hyperSlice(targetOffset, 2, index),
                                                factors2D)
                                }
                                N5Utils.saveBlock(targetOffset, sourceContainer, targetDataset, attributes, block.b)
                                null
                        }
                        es.submit(task)
                }
                futures.forEach { it.get() }
                println("Done!")
                println(targetInterval)
                es.shutdown()
                return 0
        }
}

fun N5Reader.datasetAsDoubleType(dataset: String): RandomAccessibleInterval<DoubleType> {
        return when (this.getDatasetAttributes(dataset).dataType) {
                DataType.FLOAT32 -> open<FloatType>(dataset).asDoubleType()
                DataType.FLOAT64 -> open<DoubleType>(dataset)
                DataType.INT8    -> open<ByteType>(dataset).asDoubleType()
                DataType.UINT8   -> open<UnsignedByteType>(dataset).asDoubleType()
                else             -> null as RandomAccessibleInterval<DoubleType>?
        }!!

}

fun <T: NativeType<T>> N5Reader.open(dataset: String): RandomAccessibleInterval<T> = N5Utils.open<T>(this, dataset)

fun <T: RealType<T>> RandomAccessibleInterval<T>.asDoubleType() = Converters.convert(this, RealDoubleConverter(), DoubleType())

fun N5Reader.firstMipmapLevelDataset(group: String): String {
        return if (datasetExists(group)) {
                group
        } else if (!exists(group)) {
                throw Exception("$group does not exist in $this")
        } else if (getAttribute(group, "painteraData", JsonElement::class.java) != null) {
                firstMipmapLevelDataset("$group/data")
        } else {
                firstMipmapLevelDataset("$group/s0")
        }
}

exitProcess(CommandLine(CliArgs()).execute(*args))

## repos.md

      
    Raw
  

              repos.md
            
          
dissertation on defense branch
cremi_python on qip branch
	_convert() {
	ITERATION=$1
	SETUP=$2
	THRESHOLD=$3
	SAMPLE=$4
	CONTAINER="/nrs/saalfeld/hanslovskyp/experiments/quasi-isotropic-predictions/affinities-glia/neuron_ids-unlabeled-unmask-background/predictions/CREMI/sample_${SAMPLE}.n5"
	BASE_PATH="volumes/predictions/neuron_ids-unlabeled-unmask-background/${SETUP}/${ITERATION}"
	if [ -z "${THRESHOLD}" ]; then
	SOURCE="${BASE_PATH}/mutex-watershed-merged"
	else
	SOURCE="${BASE_PATH}/mutex-watershed-threshold=${THRESHOLD}-merged"
	fi
	TARGET="${SOURCE}-multiscale"

	if [ -d "${CONTAINER}/${TARGET}" ]; then
	echo "${CONTAINER}/${TARGET} already exists. Skipping..."
	exit 0
	fi

	paintera-convert \
	to-paintera \
	--block-size=64,64,64 \
	--scale 2 2 2 2 2 \
	--output-container="${CONTAINER}" \
	--container="${CONTAINER}" \
	--dataset="${SOURCE}" \
	--overwrite-existing \
	--target-dataset="${TARGET}"
	echo "Successfully converted into dataset ${TARGET} in container ${CONTAINER}."

	}

	mkdir -p $HOME/.local/tmp/mutex-watersheds-eval/logs

	for ITERATION in 500000; do
	for SETUP in 0 3; do
	for THRESHOLD in 0.5 '' 0.3 0.7 0.1 0.9; do
	for SAMPLE in 2 1 0 A C B; do
	_convert $ITERATION $SETUP "$THRESHOLD" $SAMPLE
	done
	done
	done
	done
	_evaluate() {
	ITERATION=$1
	SETUP=$2
	SAMPLE=$3
	THRESHOLD=$4
	if [ -z "$5" ]; then
	WITH_GLIA=''
	WITH_GLIA_ARG='--with-glia=false'
	else
	WITH_GLIA='-with-glia'
	WITH_GLIA_ARG='--with-glia=true'
	fi
	if [ -z "$THRESHOLD" ]; then
	eval-mutex-watersheds \
	--sample=$SAMPLE \
	--setup=$SETUP \
	--iteration=$ITERATION \
	--border-threshold=175 \
	--overwrite-existing \
	"${WITH_GLIA_ARG}" \
	1> "$HOME/.local/tmp/mutex-watersheds-eval/logs/eval.log.sample=${SAMPLE}_setup=${SETUP}_iteration=${ITERATION}${WITH_GLIA}" \
	2>&1
	eval-mutex-watersheds \
	--sample=$SAMPLE \
	--setup=$SETUP \
	--iteration=$ITERATION \
	--border-threshold=175 \
	--ignore-training-mask \
	--overwrite-existing \
	"${WITH_GLIA_ARG}" \
	1> "$HOME/.local/tmp/mutex-watersheds-eval/logs/eval.log.sample=${SAMPLE}_setup=${SETUP}_iteration=${ITERATION}${WITH_GLIA}_ignore-training-mask" \
	2>&1
	else
	eval-mutex-watersheds \
	--sample=$SAMPLE \
	--setup=$SETUP \
	--iteration=$ITERATION \
	--threshold=$THRESHOLD \
	--border-threshold=175 \
	--overwrite-existing \
	"${WITH_GLIA_ARG}" \
	1> "$HOME/.local/tmp/mutex-watersheds-eval/logs/eval.log.sample=${SAMPLE}_setup=${SETUP}_iteration=${ITERATION}_threshold=${THRESHOLD}${WITH_GLIA}" \
	2>&1
	eval-mutex-watersheds \
	--sample=$SAMPLE \
	--setup=$SETUP \
	--iteration=$ITERATION \
	--threshold=$THRESHOLD \
	--border-threshold=175 \
	--ignore-training-mask \
	--overwrite-existing \
	"${WITH_GLIA_ARG}" \
	1> "$HOME/.local/tmp/mutex-watersheds-eval/logs/eval.log.sample=${SAMPLE}_setup=${SETUP}_iteration=${ITERATION}_threshold=${THRESHOLD}${WITH_GLIA}_ignore-training-mask" \
	2>&1
	fi
	}

	_evaluate_all_thresholds() {
	ITERATION=$1
	SETUP=$2
	SAMPLE=$3
	WITH_GLIA="$4"
	for THRESHOLD in 0.5 '' 0.3 0.7 0.1 0.9; do
	_evaluate $ITERATION $SETUP $SAMPLE "$THRESHOLD" "$WITH_GLIA"
	done

	}

	mkdir -p $HOME/.local/tmp/mutex-watersheds-eval/logs

	for ITERATION in 500000; do
	for SETUP in 0 3; do
	for SAMPLE in A B C 0 1 2; do
	_evaluate_all_thresholds $ITERATION $SETUP $SAMPLE 1 &
	done
	done
	done

	wait
	N_NODES=${N_NODES:-10}
	N_EXECUTORS_PER_NODE=${N_EXECUTORS_PER_NODE:-15}

	BLOCK_SIZE=${BLOCK_SIZE:-64,64,64}

	# N5 group that holds integer type labels
	# N5_GROUP="/nrs/saalfeld/hanslovskyp/n5-examples"
	BLOCK=${BLOCK:-03}
	CONTAINER="${CONTAINER:-/nrs/saalfeld/hanslovskyp/experiments/quasi-isotropic-predictions/affinities-glia/neuron_ids-unlabeled-unmask-background/predictions/lauritzen/${BLOCK}/workspace.n5}"
	if [ -z "${THRESHOLD}" ]; then
	SOURCE_DATASET="${SOURCE_DATASET:-volumes/predictions/neuron_ids-unlabeled-unmask-background/0/500000/mutex-watershed-merged}"
	TARGET_DATASET="${TARGET_DATASET:-${SOURCE_DATASET}-multiscale}"
	else
	SOURCE_DATASET="${SOURCE_DATASET:-volumes/predictions/neuron_ids-unlabeled-unmask-background/0/500000/mutex-watershed-threshold=${THRESHOLD}-merged}"
	TARGET_DATASET="${TARGET_DATASET:-${SOURCE_DATASET}-multiscale}"
	fi
	# input dataset
	DATASET="filtered/volumes/labels/multiset-mipmap-with-label-lists-per-block"

	JAR="${JAR:-$HOME/paintera-conversion-helper-0.9.2-shaded.jar}"
	CLASS="org.janelia.saalfeldlab.paintera.conversion.PainteraConvert"

	N_EXECUTORS_PER_NODE=$N_EXECUTORS_PER_NODE \
	$HOME/flintstone/flintstone.sh ${N_NODES} $JAR $CLASS \
	to-paintera \
	--block-size=$BLOCK_SIZE \
	--scale 2 2 2 2 2 2 2 2 \
	-m -1 -1 -1
	5 4 3 2 \
	--output-container="${CONTAINER}" \
	--container="${CONTAINER}" \
	--dataset="${SOURCE_DATASET}" \
	--target-dataset="${TARGET_DATASET}"
	BASE_PATH="/nrs/saalfeld/hanslovskyp/experiments/quasi-isotropic-predictions/affinities-glia/neuron_ids-unlabeled-unmask-background/predictions/CREMI"
	SAMPLE="$1"
	PREDICTION_CONTAINER="${BASE_PATH}/sample_${SAMPLE}.n5"
	SETUP="$2"
	THRESHOLD="$3"
	DATASET_BASE="volumes/predictions/neuron_ids-unlabeled-unmask-background/$SETUP/500000"
	if [ -z "$THRESHOLD" ]; then
	MERGED_STRING="mutex-watershed-merged-multiscale"
	WITH_GLIA_STRING="mutex-watershed-merged-with-glia"
	else
	MERGED_STRING="mutex-watershed-threshold=$THRESHOLD-merged-multiscale"
	WITH_GLIA_STRING="mutex-watershed-threshold=$THRESHOLD-merged-with-glia"
	fi
	paintera \
	--add-n5-container "/groups/saalfeld/home/hanslovskyp/data/from-arlo/interpolated-combined/sample_${SAMPLE}.n5" --resolution=36,36,360 -d volumes/raw \
	--add-n5-container "/groups/saalfeld/home/hanslovskyp/data/from-arlo/interpolated-combined/sample_${SAMPLE}.n5" --revert-array-attributes -d volumes/labels/glia-downsampled volumes/labels/neuron_ids-downsampled \
	--add-n5-container "${PREDICTION_CONTAINER}" -d "${DATASET_BASE}/affinities" "${DATASET_BASE}/glia" "${DATASET_BASE}/${MERGED_STRING}" "${DATASET_BASE}/${WITH_GLIA_STRING}" --channels 8,4,0 9,5,1 10,6,2 11,7,3
	#!/usr/bin/env kscript

	@file:KotlinOpts("-J-Xmx100g")

	@file:MavenRepository("scijava.public", "https://maven.scijava.org/content/groups/public")

	@file:DependsOn("org.janelia.saalfeldlab:n5:2.1.2")
	@file:DependsOn("org.janelia.saalfeldlab:n5-imglib2:3.4.1")
	@file:DependsOn("org.janelia.saalfeldlab:k5:0.1.0")
	@file:DependsOn("org.apache.commons:commons-compress:1.18")
	@file:DependsOn("info.picocli:picocli:4.0.2")
	@file:DependsOn("net.imglib2:imglib2-algorithm:0.11.1")
	@file:DependsOn("net.imglib2:imglib2-realtransform:2.2.1")
	// use logback to set log level programmatically
	@file:DependsOn("ch.qos.logback:logback-classic:1.3.0-alpha4")
	@file:DependsOn("sc.fiji:bigdataviewer-core:7.0.0")
	// @file:DependsOn("org.slf4j:slf4j-api:2.0.0-alpha0")
	// @file:DependsOn("org.slf4j:slf4j-simple:2.0.0-alpha0")


	import java.io.File
	import java.lang.Runtime
	import java.lang.Thread
	import java.util.concurrent.Callable
	import java.util.concurrent.Executors

	import kotlin.system.exitProcess

	import bdv.export.Downsample
	import com.google.gson.JsonElement
	import net.imglib2.FinalInterval
	import net.imglib2.RandomAccessibleInterval
	import net.imglib2.algorithm.gauss3.Gauss3
	import net.imglib2.algorithm.util.Grids
	import net.imglib2.converter.Converter
	import net.imglib2.converter.Converters
	import net.imglib2.converter.RealDoubleConverter
	import net.imglib2.img.array.ArrayImgs
	import net.imglib2.interpolation.randomaccess.NLinearInterpolatorFactory
	import net.imglib2.realtransform.AffineTransform3D
	import net.imglib2.realtransform.RealViews
	import net.imglib2.type.NativeType
	import net.imglib2.type.numeric.RealType
	import net.imglib2.type.numeric.integer.ByteType
	import net.imglib2.type.numeric.integer.UnsignedByteType
	import net.imglib2.type.numeric.real.DoubleType
	import net.imglib2.type.numeric.real.FloatType
	import net.imglib2.util.Fraction
	import net.imglib2.util.Intervals
	import net.imglib2.view.Views
	import org.janelia.saalfeldlab.k5.K5
	import org.janelia.saalfeldlab.n5.DatasetAttributes
	import org.janelia.saalfeldlab.n5.DataType
	import org.janelia.saalfeldlab.n5.GzipCompression
	import org.janelia.saalfeldlab.n5.N5FSReader
	import org.janelia.saalfeldlab.n5.N5FSWriter
	import org.janelia.saalfeldlab.n5.N5Reader
	import org.janelia.saalfeldlab.n5.imglib2.N5Utils
	import picocli.CommandLine

	// This is necessary to get auto-detection of n5 compression to work, for some reason.
	Thread
	.currentThread()
	.setContextClassLoader(java.lang.invoke.MethodHandles.lookup().lookupClass().getClassLoader())

	@CommandLine.Command(name = "qi-fy")
	class CliArgs : Callable<Int> {

	@CommandLine.Option(names = ["--container", "-c"], required = true)
	lateinit var containerPath: File

	@CommandLine.Option(names = ["--dataset", "-d"], required = true)
	lateinit var dataset: String

	@CommandLine.Option(names = ["--block-size"], required = false, defaultValue = "64", split = ",")
	lateinit var blockSize: IntArray

	@CommandLine.Option(names = ["--resolution"], required = false, defaultValue = "108,108,120", split = ",")
	lateinit var resolution: DoubleArray

	@CommandLine.Option(names = ["--offset"], required = false, defaultValue = "0,0,0", split = ",")
	lateinit var offset: DoubleArray

	@CommandLine.Option(names = ["--parallelism", "-p"], required = false)
	var parallelism: Int? = null

	override fun call(): Int {
	val blockSize = if (this.blockSize.size == 3) blockSize else IntArray(3) { this.blockSize[0] }
	val sourceContainer = with (K5) { containerPath.absolutePath.n5fsWriter() }
	val sourceDataset = sourceContainer.firstMipmapLevelDataset(dataset)
	val targetDataset = "$dataset-qified"
	val scaleFactorsTargetToSource = arrayOf(Fraction(3, 1), Fraction(3, 1), Fraction(1, 3))
	val offsetsInOriginalCoordinates = arrayOf(Fraction(1, 1), Fraction(1, 1), Fraction(0, 1))
	val translationTargetToSource = offsetsInOriginalCoordinates.map { it.ratio }.toDoubleArray()
	val scaleTargetToSource = scaleFactorsTargetToSource.map { it.ratio }.toDoubleArray()
	val scaleFactorsSourceToTarget = scaleFactorsTargetToSource.map { it.clone().also { it.invert() } }
	val source = sourceContainer.datasetAsDoubleType(sourceDataset)
	println(LongArray(3) { source.dimension(it) }.map { it })
	val targetSize = LongArray(3) { (source.dimension(it) * scaleFactorsSourceToTarget[it].ratio).toLong() }
	targetSize[2] -= 2L
	println(targetSize.map { it })
	val targetInterval = FinalInterval(*targetSize)
	val extendedSource = Views.extendBorder(source)
	val interpolatedSource = Views.interpolate(extendedSource, NLinearInterpolatorFactory<DoubleType>())
	val relevantTransform = AffineTransform3D()
	// need to shift input by 1 in x and y and scale by a factor of 3 along z
	relevantTransform.set(
	1.0, 0.0, 0.0, 1.0,
	0.0, 1.0, 0.0, 1.0,
	0.0, 0.0, 3.0, 0.0)
	val transformedSource = RealViews.affine(interpolatedSource, relevantTransform)

	val attributes = DatasetAttributes(
	targetSize,
	blockSize,
	DataType.FLOAT64,
	GzipCompression())
	sourceContainer.createDataset(targetDataset, attributes)
	sourceContainer.setAttribute(targetDataset, "resolution", resolution)
	sourceContainer.setAttribute(targetDataset, "offset", offset)
	val parallelism = this.parallelism ?: Runtime.getRuntime().availableProcessors()
	val es = Executors.newFixedThreadPool(parallelism)

	val factors2D = IntArray(2) { scaleFactorsTargetToSource[it].ratio.toInt() }

	val intervals = Grids.collectAllContainedIntervalsWithGridPositions(targetSize, blockSize)
	val futures = intervals.map { block ->
	val task = Callable<Unit> {
	val target = ArrayImgs.doubles(*Intervals.dimensionsAsLongArray(block.a))
	val targetOffset = Views.translate(target, *Intervals.minAsLongArray(block.a))
	println("doing stuff for $targetOffset")
	for (index in targetOffset.min(2) .. targetOffset.max(2)) {
	Downsample.downsample(
	Views.hyperSlice(transformedSource, 2, index),
	Views.hyperSlice(targetOffset, 2, index),
	factors2D)
	}
	N5Utils.saveBlock(targetOffset, sourceContainer, targetDataset, attributes, block.b)
	null
	}
	es.submit(task)
	}
	futures.forEach { it.get() }
	println("Done!")
	println(targetInterval)
	es.shutdown()
	return 0
	}
	}

	fun N5Reader.datasetAsDoubleType(dataset: String): RandomAccessibleInterval<DoubleType> {
	return when (this.getDatasetAttributes(dataset).dataType) {
	DataType.FLOAT32 -> open<FloatType>(dataset).asDoubleType()
	DataType.FLOAT64 -> open<DoubleType>(dataset)
	DataType.INT8 -> open<ByteType>(dataset).asDoubleType()
	DataType.UINT8 -> open<UnsignedByteType>(dataset).asDoubleType()
	else -> null as RandomAccessibleInterval<DoubleType>?
	}!!

	}

	fun <T: NativeType<T>> N5Reader.open(dataset: String): RandomAccessibleInterval<T> = N5Utils.open<T>(this, dataset)

	fun <T: RealType<T>> RandomAccessibleInterval<T>.asDoubleType() = Converters.convert(this, RealDoubleConverter(), DoubleType())

	fun N5Reader.firstMipmapLevelDataset(group: String): String {
	return if (datasetExists(group)) {
	group
	} else if (!exists(group)) {
	throw Exception("$group does not exist in $this")
	} else if (getAttribute(group, "painteraData", JsonElement::class.java) != null) {
	firstMipmapLevelDataset("$group/data")
	} else {
	firstMipmapLevelDataset("$group/s0")
	}
	}

	exitProcess(CommandLine(CliArgs()).execute(*args))