align_star.config

process {

    // STAR alignment configuration
    // Conditionals for when these processes run are handled in the workflow

    withName: '.*ALIGN_STAR:STAR_ALIGN|.*ALIGN_STAR:STAR_ALIGN_IGENOMES|.*ALIGN_STAR:SENTIEON_STAR_ALIGN|.*ALIGN_STAR:PARABRICKS_RNA_FQ2BAM' {
        ext.args = {
            def isPbrun = task.process.contains('PARABRICKS')
            def quantifier = params.aligner == 'star_rsem' ? 'rsem' : 'salmon'

            // iGenomes is pinned to STAR 2.6.1d which uses the legacy --quantTranscriptomeBan flag.
            // Sentieon also uses this legacy flag for now.
            // If Sentieon updates to support the STAR 2.7+ flag, remove it from this check.
            def useLegacyBanFlag = task.process.endsWith(':STAR_ALIGN_IGENOMES') || task.process.endsWith(':SENTIEON_STAR_ALIGN')

            def args = []

            // Common args - pbrun uses kebab-case equivalents
            if (isPbrun) {
                args += [
                    '--quantMode TranscriptomeSAM',
                    '--out-sam-attributes NH HI AS NM MD',
                    '--read-files-command zcat',
                    "--read-group-sm ${task.tag}",
                    "--read-group-id-prefix ${task.tag}"
                ]
            } else {
                args += [
                    '--quantMode TranscriptomeSAM',
                    '--outSAMtype BAM Unsorted',
                    '--outSAMattributes NH HI AS NM MD',
                    '--readFilesCommand zcat'
                ]
            }

            // Quantifier-specific args
            //
            // NOTE: pbrun rna_fq2bam is based on STAR 2.7.2a and does NOT support:
            //   --outFilterType BySJout  : no pbrun equivalent; pbrun uses its own filtering defaults
            //   --sjdbScore 1            : no pbrun equivalent; affects junction scoring priority
            //   --quantTranscriptomeBan Singleend / --quantTranscriptomeSAMoutput BanSingleEnd :
            //       no pbrun equivalent; Salmon handles mixed single/paired records gracefully
            //   --runRNGseed 0           : no pbrun equivalent; pbrun uses deterministic alignment selection
            //
            if (quantifier == 'rsem') {
                if (isPbrun) {
                    args += [
                        '--out-sam-unmapped Within',
                        '--max-out-filter-multimap 20',
                        '--max-out-filter-mismatch 999',
                        '--max-out-filter-mismatch-ratio 0.04',
                        '--min-intron-size 20',
                        '--max-intron-size 1000000',
                        '--max-align-mates-gap 1000000',
                        '--min-align-sj-overhang 8',
                        '--min-align-sjdb-overhang 1'
                    ]
                } else {
                    args += [
                        '--outSAMunmapped Within',
                        '--outFilterType BySJout',
                        '--outFilterMultimapNmax 20',
                        '--outFilterMismatchNmax 999',
                        '--outFilterMismatchNoverLmax 0.04',
                        '--alignIntronMin 20',
                        '--alignIntronMax 1000000',
                        '--alignMatesGapMax 1000000',
                        '--alignSJoverhangMin 8',
                        '--alignSJDBoverhangMin 1',
                        '--sjdbScore 1'
                    ]
                }
            } else {
                if (isPbrun) {
                    args += [
                        '--two-pass-mode Basic',
                        '--max-out-filter-multimap 20',
                        '--min-align-sjdb-overhang 1',
                        '--out-sam-strand-field intronMotif'
                    ]
                } else {
                    args += [
                        '--twopassMode Basic',
                        '--runRNGseed 0',
                        '--outFilterMultimapNmax 20',
                        '--alignSJDBoverhangMin 1',
                        '--outSAMstrandField intronMotif',
                        useLegacyBanFlag ? '--quantTranscriptomeBan Singleend' : '--quantTranscriptomeSAMoutput BanSingleEnd'
                    ]
                }
            }

            // Unmapped reads output
            if (params.save_unaligned || (params.contaminant_screening && params.contaminant_screening_input == 'unmapped')) {
                args += isPbrun
                    ? ['--out-reads-unmapped Fastx']
                    : ['--outReadsUnmapped Fastx']
            }

            // Prokaryotic: disable spliced alignment
            if (params.prokaryotic) {
                if (isPbrun) {
                    args += ['--max-intron-size 1']
                } else {
                    args += ['--sjdbGTFfeatureExon CDS', '--alignIntronMax 1']
                }
            }

            // Merge user-supplied extra args, deduplicating so user values
            // override pipeline defaults (prevents STAR "duplicate parameter" error)
            if (params.extra_star_align_args) {
                def baseMap = args.join(' ').split(/\s(?=--)/).collectEntries { String token ->
                    def parts = token.trim().split(/\s+/, 2)
                    [(parts[0]): parts.size() > 1 ? parts[1] : '']
                }
                def extraMap = params.extra_star_align_args.split(/\s(?=--)/).collectEntries { String token ->
                    def parts = token.trim().split(/\s+/, 2)
                    [(parts[0]): parts.size() > 1 ? parts[1] : '']
                }
                args = (baseMap + extraMap).collect { k, v -> v ? "${k} ${v}" : k }
            }

            // Read group tags - assembled once from samplesheet meta and pipeline params
            // Per-sample meta values take priority over global params
            def seqCenter   = meta.seq_center   ?: params.seq_center
            def seqPlatform = meta.seq_platform ?: params.seq_platform
            if (isPbrun) {
                // pbrun supports --read-group-pl but has no --read-group-cn flag
                if (seqPlatform) args << "--read-group-pl ${seqPlatform}"
            } else {
                def rgParts = ["'ID:${meta.id}'", "'SM:${meta.id}'"]
                if (seqPlatform) rgParts << "'PL:${seqPlatform}'"
                if (seqCenter) rgParts << "'CN:${seqCenter}'"
                args += ["--outSAMattrRGline ${rgParts.join(' ')}"]
            }

            args.join(' ')
        }
    }

    withName: '.*ALIGN_STAR:STAR_ALIGN|.*ALIGN_STAR:STAR_ALIGN_IGENOMES|.*ALIGN_STAR:SENTIEON_STAR_ALIGN|.*ALIGN_STAR:PARABRICKS_RNA_FQ2BAM' {
        publishDir = [
            [
                path: { "${params.outdir}/${task.ext.publish_prefix}${params.aligner}/log" },
                mode: params.publish_dir_mode,
                pattern: '*.{out,tab}'
            ],
            [
                path: { params.save_align_intermeds ? "${params.outdir}/${task.ext.publish_prefix}${params.aligner}" : params.outdir },
                mode: params.publish_dir_mode,
                pattern: '*.bam',
                saveAs: { filename -> params.save_align_intermeds ? filename : null }
            ],
            [
                path: { params.save_unaligned ? "${params.outdir}/${task.ext.publish_prefix}${params.aligner}/unmapped" : params.outdir },
                mode: params.publish_dir_mode,
                pattern: '*.fastq.gz',
                saveAs: { filename -> params.save_unaligned ? filename : null }
            ]
        ]
    }

}