Recombination Section Enhancements

build.gradle.kts

@@ -4,7 +4,7 @@ plugins {
 }
 
 group = "net.maizegenetics"
-version = "0.2.6"
+version = "0.2.7"
 
 repositories {
     mavenCentral()

src/main/kotlin/net/maizegenetics/commands/AlignMutatedAssemblies.kt

@@ -19,8 +19,8 @@ import kotlin.system.exitProcess
 
 class AlignMutatedAssemblies : CliktCommand(name = "align-mutated-assemblies") {
     companion object {
-        private const val LOG_FILE_NAME = "05_align_mutated_assemblies.log"
-        private const val MUTATED_ALIGNMENT_RESULTS_DIR = "05_mutated_alignment_results"
+        private const val LOG_FILE_NAME = "10_align_mutated_assemblies.log"
+        private const val MUTATED_ALIGNMENT_RESULTS_DIR = "10_mutated_alignment_results"
         private const val MAF_PATHS_FILE = "maf_file_paths.txt"
 
         // minimap2 parameters
@@ -71,7 +71,7 @@ class AlignMutatedAssemblies : CliktCommand(name = "align-mutated-assemblies") {
 
     private val outputDir by option(
         "--output-dir", "-o",
-        help = "Custom output directory (default: work_dir/output/05_mutated_alignment_results)"
+        help = "Custom output directory (default: work_dir/output/10_mutated_alignment_results)"
     ).path(mustExist = false, canBeFile = false, canBeDir = true)
 
     private fun collectFastaFiles(): List<Path> {

src/main/kotlin/net/maizegenetics/commands/ConvertCoordinates.kt

@@ -18,12 +18,12 @@ import kotlin.system.exitProcess
 
 class ConvertCoordinates : CliktCommand(name = "convert-coordinates") {
     companion object {
-        private const val LOG_FILE_NAME = "08_convert_coordinates.log"
-        private const val COORDS_RESULTS_DIR = "08_coordinates_results"
+        private const val LOG_FILE_NAME = "07_convert_coordinates.log"
+        private const val COORDS_RESULTS_DIR = "07_coordinates_results"
         private const val KEY_PATHS_FILE = "key_file_paths.txt"
         private const val FOUNDER_KEY_PATHS_FILE = "founder_key_file_paths.txt"
         private const val PYTHON_SCRIPT = "src/python/cross/convert_coords.py"
-        private const val DEFAULT_REFKEY_DIR = "06_crossovers_results"
+        private const val DEFAULT_REFKEY_DIR = "05_crossovers_results"
     }
 
     private val logger: Logger = LogManager.getLogger(ConvertCoordinates::class.java)
@@ -53,7 +53,7 @@ class ConvertCoordinates : CliktCommand(name = "convert-coordinates") {
 
     private val outputDirOption by option(
         "--output-dir", "-o",
-        help = "Custom output directory (default: work_dir/output/08_coordinates_results)"
+        help = "Custom output directory (default: work_dir/output/07_coordinates_results)"
     ).path(mustExist = false, canBeFile = false, canBeDir = true)
 
     override fun run() {
@@ -112,14 +112,16 @@ class ConvertCoordinates : CliktCommand(name = "convert-coordinates") {
 
         // Run convert_coords.py
         // Set PYTHONPATH so Python can find the 'python' package for imports
-        val pythonPath = mlimputeDir.resolve("src").toString()
+        // Use absolute paths since the working directory is set to outputDir
+        // Use sh -c to set PYTHONPATH inside pixi's environment
+        val pythonPath = mlimputeDir.resolve("src").toAbsolutePath().toString()
+        val scriptPath = pythonScript.toAbsolutePath().toString()
+        val assemblyListPath = assemblyList.toAbsolutePath().toString()
+        val chainDirPath = chainDir.toAbsolutePath().toString()
+        val shellCommand = "PYTHONPATH='$pythonPath' python '$scriptPath' --assembly-list '$assemblyListPath' --chain-dir '$chainDirPath'"
         logger.info("Running convert_coords.py")
         val exitCode = ProcessRunner.runCommand(
-            "env", "PYTHONPATH=$pythonPath",
-            "pixi", "run",
-            "python", pythonScript.toString(),
-            "--assembly-list", assemblyList.toString(),
-            "--chain-dir", chainDir.toString(),
+            "pixi", "run", "sh", "-c", shellCommand,
             workingDir = outputDir.toFile(),
             logger = logger
         )

src/main/kotlin/net/maizegenetics/commands/CreateChainFiles.kt

@@ -19,8 +19,8 @@ import kotlin.system.exitProcess
 
 class CreateChainFiles : CliktCommand(name = "create-chain-files") {
     companion object {
-        private const val LOG_FILE_NAME = "07_create_chain_files.log"
-        private const val CHAIN_RESULTS_DIR = "07_chain_results"
+        private const val LOG_FILE_NAME = "06_create_chain_files.log"
+        private const val CHAIN_RESULTS_DIR = "06_chain_results"
         private const val CHAIN_PATHS_FILE = "chain_file_paths.txt"
         private const val BASH_SCRIPT = "src/python/cross/create_chains.sh"
         private const val DEFAULT_JOBS = 8
@@ -49,7 +49,7 @@ class CreateChainFiles : CliktCommand(name = "create-chain-files") {
 
     private val outputDirOption by option(
         "--output-dir", "-o",
-        help = "Custom output directory (default: work_dir/output/07_chain_results)"
+        help = "Custom output directory (default: work_dir/output/06_chain_results)"
     ).path(mustExist = false, canBeFile = false, canBeDir = true)
 
     private fun collectMafFiles(): List<Path> {
@@ -128,20 +128,34 @@ class CreateChainFiles : CliktCommand(name = "create-chain-files") {
         logger.info("Cleaning up temporary MAF directory")
         tempMafDir.toFile().deleteRecursively()
 
-        // Collect generated chain files
+        // Collect generated chain files (all files with .chain extension)
         val chainFiles = outputDir.listDirectoryEntries()
             .filter { it.isRegularFile() && it.extension == "chain" }
             .sorted()
 
-        if (chainFiles.isEmpty()) {
+        // Rename chain files to include "_subsampled" suffix if not already present
+        val renamedChainFiles = chainFiles.map { chainFile ->
+            val fileName = chainFile.nameWithoutExtension
+            if (!fileName.endsWith("_subsampled")) {
+                val newFileName = "${fileName}_subsampled.chain"
+                val newPath = chainFile.parent.resolve(newFileName)
+                chainFile.moveTo(newPath, overwrite = true)
+                logger.info("Renamed ${chainFile.fileName} to $newFileName")
+                newPath
+            } else {
+                chainFile
+            }
+        }.sorted()
+
+        if (renamedChainFiles.isEmpty()) {
             logger.warn("No chain files generated")
         } else {
-            logger.info("Generated ${chainFiles.size} chain file(s):")
-            chainFiles.forEach { logger.info("  $it") }
+            logger.info("Generated ${renamedChainFiles.size} chain file(s):")
+            renamedChainFiles.forEach { logger.info("  $it") }
 
             // Write chain file paths to text file
             FileUtils.writeFilePaths(
-                chainFiles,
+                renamedChainFiles,
                 outputDir.resolve(CHAIN_PATHS_FILE),
                 logger,
                 "Chain file"

src/main/kotlin/net/maizegenetics/commands/FormatRecombinedFastas.kt

@@ -18,12 +18,12 @@ import kotlin.system.exitProcess
 
 class FormatRecombinedFastas : CliktCommand(name = "format-recombined-fastas") {
     companion object {
-        private const val LOG_FILE_NAME = "10_format_recombined_fastas.log"
-        private const val FORMATTED_RESULTS_DIR = "10_formatted_fastas"
+        private const val LOG_FILE_NAME = "09_format_recombined_fastas.log"
+        private const val FORMATTED_RESULTS_DIR = "09_formatted_fastas"
         private const val FORMATTED_FASTA_PATHS_FILE = "formatted_fasta_paths.txt"
         private const val DEFAULT_LINE_WIDTH = 60
         private const val DEFAULT_THREADS = 8
-        private const val DEFAULT_INPUT_DIR = "09_recombined_sequences/recombinate_fastas"
+        private const val DEFAULT_INPUT_DIR = "08_recombined_sequences/recombinate_fastas"
     }
 
     private val logger: Logger = LogManager.getLogger(FormatRecombinedFastas::class.java)
@@ -53,7 +53,7 @@ class FormatRecombinedFastas : CliktCommand(name = "format-recombined-fastas") {
 
     private val outputDirOption by option(
         "--output-dir", "-o",
-        help = "Custom output directory (default: work_dir/output/10_formatted_fastas)"
+        help = "Custom output directory (default: work_dir/output/09_formatted_fastas)"
     ).path(mustExist = false, canBeFile = false, canBeDir = true)
 
     private fun collectFastaFiles(): List<Path> {

src/main/kotlin/net/maizegenetics/commands/GenerateRecombinedSequences.kt

@@ -12,18 +12,19 @@ import net.maizegenetics.utils.ProcessRunner
 import net.maizegenetics.utils.ValidationUtils
 import org.apache.logging.log4j.LogManager
 import org.apache.logging.log4j.Logger
+import java.nio.file.Files
 import java.nio.file.Path
 import kotlin.io.path.*
 import kotlin.system.exitProcess
 
 class GenerateRecombinedSequences : CliktCommand(name = "generate-recombined-sequences") {
     companion object {
-        private const val LOG_FILE_NAME = "09_generate_recombined_sequences.log"
-        private const val RECOMBINED_RESULTS_DIR = "09_recombined_sequences"
+        private const val LOG_FILE_NAME = "08_generate_recombined_sequences.log"
+        private const val RECOMBINED_RESULTS_DIR = "08_recombined_sequences"
         private const val RECOMBINED_FASTAS_DIR = "recombinate_fastas"
         private const val FASTA_PATHS_FILE = "recombined_fasta_paths.txt"
         private const val PYTHON_SCRIPT = "src/python/cross/write_fastas.py"
-        private const val DEFAULT_FOUNDER_KEY_DIR = "08_coordinates_results"
+        private const val DEFAULT_FOUNDER_KEY_DIR = "07_coordinates_results"
     }
 
     private val logger: Logger = LogManager.getLogger(GenerateRecombinedSequences::class.java)
@@ -59,7 +60,7 @@ class GenerateRecombinedSequences : CliktCommand(name = "generate-recombined-seq
 
     private val outputDirOption by option(
         "--output-dir", "-o",
-        help = "Custom output directory (default: work_dir/output/09_recombined_sequences)"
+        help = "Custom output directory (default: work_dir/output/08_recombined_sequences)"
     ).path(mustExist = false, canBeFile = false, canBeDir = true)
 
     override fun run() {
@@ -119,17 +120,58 @@ class GenerateRecombinedSequences : CliktCommand(name = "generate-recombined-seq
             }
         }
 
+        // Create symlinks for FASTA files with different extensions
+        // The Python script expects .fa extension, but files might have .fasta or .fna
+        logger.info("Checking FASTA file extensions in assembly directory")
+        val assemblyNames = assemblyList.readLines()
+            .filter { it.isNotBlank() }
+            .mapNotNull { line ->
+                val parts = line.split("\t")
+                if (parts.size >= 2) parts[1].trim() else null
+            }
+
+        assemblyNames.forEach { name ->
+            val faFile = assemblyDir.resolve("$name.fa")
+            if (!faFile.exists()) {
+                // Look for alternative extensions
+                val alternatives = listOf("fasta", "fna")
+                for (ext in alternatives) {
+                    val altFile = assemblyDir.resolve("$name.$ext")
+                    if (altFile.exists()) {
+                        // Create symlink: name.fa -> name.fasta (or .fna)
+                        try {
+                            Files.createSymbolicLink(faFile, altFile.fileName)
+                            logger.debug("Created symlink: ${faFile.fileName} -> ${altFile.fileName}")
+
+                            // Also create symlink for the FASTA index file if it exists
+                            val altIndexFile = assemblyDir.resolve("$name.$ext.fai")
+                            val faIndexFile = assemblyDir.resolve("$name.fa.fai")
+                            if (altIndexFile.exists() && !faIndexFile.exists()) {
+                                Files.createSymbolicLink(faIndexFile, altIndexFile.fileName)
+                                logger.debug("Created symlink: ${faIndexFile.fileName} -> ${altIndexFile.fileName}")
+                            }
+                        } catch (e: Exception) {
+                            logger.warn("Failed to create symlink for $name: ${e.message}")
+                        }
+                        break
+                    }
+                }
+            }
+        }
+
         // Run write_fastas.py
         // Set PYTHONPATH so Python can find the 'python' package for imports
-        val pythonPath = mlimputeDir.resolve("src").toString()
+        // Use absolute paths since the working directory is set to outputDir
+        // Use sh -c to set PYTHONPATH inside pixi's environment
+        val pythonPath = mlimputeDir.resolve("src").toAbsolutePath().toString()
+        val scriptPath = pythonScript.toAbsolutePath().toString()
+        val assemblyListPath = assemblyList.toAbsolutePath().toString()
+        val chromosomeListPath = chromosomeList.toAbsolutePath().toString()
+        val assemblyDirPath = assemblyDir.toAbsolutePath().toString()
+        val shellCommand = "PYTHONPATH='$pythonPath' python '$scriptPath' --assembly-list '$assemblyListPath' --chromosome-list '$chromosomeListPath' --assembly-dir '$assemblyDirPath'"
         logger.info("Running write_fastas.py")
         val exitCode = ProcessRunner.runCommand(
-            "env", "PYTHONPATH=$pythonPath",
-            "pixi", "run",
-            "python", pythonScript.toString(),
-            "--assembly-list", assemblyList.toString(),
-            "--chromosome-list", chromosomeList.toString(),
-            "--assembly-dir", assemblyDir.toString(),
+            "pixi", "run", "sh", "-c", shellCommand,
             workingDir = outputDir.toFile(),
             logger = logger
         )