Add SMARTer smRNA kit support and standalone RNA-seq summary reports

.gitattributes

@@ -0,0 +1,4 @@
+# Keep text byte-for-byte reproducible across Linux and Windows checkouts.
+* text=auto eol=lf
+*.png binary
+*.gz binary

.gitignore

@@ -41,3 +41,7 @@ Step_03_UMIAdapterTrimR1_old_v1.py
 # Backup files
 *.bak*
 *.bak_*
+
+# Local configuration and sample results
+EVscope.local.conf
+*_EVscope_output/

EVscope.conf

@@ -1,6 +1,6 @@
 # ==============================================================================
 # CONFIGURATION FILE FOR THE EVSCOPE PIPELINE
-# Version: 1.0.0
+# Version: 1.1.0
 # ==============================================================================
 
 # --- A. Core Paths ---
@@ -37,7 +37,60 @@ RSEM_BOWTIE2_INDEX="${EVscope_PATH}/references/index/RSEM_bowtie2_index_3659642_
 STAR_INDEX="${EVscope_PATH}/references/index/Index_STAR-2.7.10b_sjdbOverhang99_GeneCode45"
 BWA_INDEX="${EVscope_PATH}/references/index/Index_BWA_V0.7.18/hg38.p14.whole.genome.fa"
 
-# --- F. General Annotation ---
+# --- F. STAR Alignment Parameters ---
+# Bash arrays are used so every option/value remains a separate command-line token.
+# Edit the appropriate profile to change an existing option. Add new, non-duplicate
+# options to STAR_EXTRA_ARGS or the kit-specific EXTRA array.
+STAR_COMMON_ARGS=(
+    --readFilesCommand zcat
+    --outFilterMultimapNmax 100
+    --winAnchorMultimapNmax 100
+    --outSAMtype BAM SortedByCoordinate
+)
+STAR_TOTAL_RNA_ARGS=(
+    --twopassMode Basic
+    --chimSegmentMin 10
+    --chimJunctionOverhangMin 10
+    --chimScoreMin 1
+    --chimOutType Junctions WithinBAM
+)
+STAR_SMRNA_SE_ARGS=(
+    --twopassMode None
+    --alignEndsType EndToEnd
+    --alignIntronMax 1
+    --outFilterMismatchNmax 1
+    --outFilterMultimapScoreRange 0
+    --outFilterScoreMinOverLread 0
+    --outFilterMatchNminOverLread 0
+    --outFilterScoreMin 12
+    --outFilterMatchNmin 14
+    --chimSegmentMin 10
+    --chimJunctionOverhangMin 10
+    --chimScoreMin 1
+    --chimOutType Junctions WithinBAM
+)
+STAR_SMRNA_PE_ARGS=(
+    --twopassMode None
+    --alignEndsType EndToEnd
+    --alignIntronMax 1
+    --outFilterMismatchNmax 1
+    --outFilterMultimapScoreRange 0
+    --outFilterScoreMinOverLread 0
+    --outFilterMatchNminOverLread 0
+    --outFilterScoreMin 24
+    --outFilterMatchNmin 28
+    --chimSegmentMin 10
+    --chimJunctionOverhangMin 10
+    --chimScoreMin 1
+    --chimOutType Junctions WithinBAM
+)
+# Extra options applied to every STAR run, or only to the selected kit.
+# Example: STAR_SMRNA_EXTRA_ARGS=(--limitBAMsortRAM 30000000000)
+STAR_EXTRA_ARGS=()
+STAR_TOTAL_RNA_EXTRA_ARGS=()
+STAR_SMRNA_EXTRA_ARGS=()
+
+# --- G. General Annotation ---
 PIPELINE_BIN_DIR="${EVscope_PATH}/bin"
 TOTAL_GENE_GTF="${EVscope_PATH}/references/annotations_HG38/HG38_3659642_combined_RNAs_with_gold_standard_piRNAs.gtf"
 TOTAL_GENE_REFFLAT="${EVscope_PATH}/references/annotations_HG38/HG38_3659642_combined_RNAs_with_gold_standard_piRNAs.refflat"
@@ -48,7 +101,7 @@ GENCODE_V45_non_overlapping_exon_BED="${EVscope_PATH}/references/annotations_HG3
 HUMAN_RRNA_INTERVAL="${EVscope_PATH}/references/annotations_HG38/GeneCodeV45_combined_Nucl_Mt_rRNA_589.interval_list"
 ENCODE_BLACKLIST_BED="${EVscope_PATH}/references/annotations_HG38/Encode_hg38-blacklist.v2.bed"
 
-# --- G. Step 18: Merged SAF for meta-gene region counting ---
+# --- H. Step 18: Merged SAF for meta-gene region counting ---
 # Single merged SAF with 7 mutually exclusive regions (801,166 features).
 # GeneID format: {region}__region_{N}  (e.g. exon__region_1, intron__region_42)
 # Regions: 5UTR, 3UTR, exon, intron, promoter, downstream, intergenic
@@ -57,7 +110,7 @@ export STEP18_MERGED_SAF="${EVscope_PATH}/references/annotations_HG38/HG38_all7_
 # Region labels for output (order matters for pie chart)
 export STEP18_REGION_LABELS="5UTR,3UTR,exon,intron,promoter,downstream,intergenic"
 
-# --- H. Step 26a: BED files for RNA Type Density Plot ---
+# --- I. Step 26a: BED files for RNA Type Density Plot ---
 export STEP26_RNATYPE_BEDS=(
     "${EVscope_PATH}/references/annotations_HG38/HG38_protein_coding.bed"
     "${EVscope_PATH}/references/annotations_HG38/HG38_tRNAs.bed"
@@ -81,7 +134,7 @@ export STEP26_RNATYPE_BEDS=(
 )
 export STEP26_RNATYPE_LABELS="[protein_coding,tRNAs,rRNAs,Y_RNAs,snoRNAs,snRNAs,scaRNAs,vault_RNAs,lncRNAs,TEC_protein_coding,IG_genes,TR_genes,misc-sncRNAs,miRNAs,pseudogenes,ERVs,LINEs,SINEs,piRNAs]"
 
-# --- I. Step 26b: BED files for Meta-Gene Density Plot ---
+# --- J. Step 26b: BED files for Meta-Gene Density Plot ---
 export STEP26_METAGENE_BEDS=(
     "${EVscope_PATH}/references/annotations_HG38/HG38_promoter_upstream2000bp_noOverlap.bed"
     "${EVscope_PATH}/references/annotations_HG38/HG38_5UTR_noOverlap.bed"
@@ -93,7 +146,7 @@ export STEP26_METAGENE_BEDS=(
 )
 export STEP26_METAGENE_LABELS="[Promoters,5'UTR,Exon,Intron,3'UTR,Downstream,Intergenic]"
 
-# --- J. Deconvolution Refs ---
+# --- K. Deconvolution Refs ---
 GTEX_SMTS_REF="${EVscope_PATH}/references/deconvolution_HG38/human_ref_GTEx_v10_SMTS_AVG_TPM.csv"
 Monaco2020_ImmuneCell_REF="${EVscope_PATH}/references/deconvolution_HG38/human_ref_Monaco2020_ImmuneCell_AVG_TPM.csv"
 BRAIN_SC_REF="${EVscope_PATH}/references/deconvolution_HG38/human_ref_brain_single_cell_atlasV1_31superclusters_AVP_CPM.csv"