1. Splitting large tables into 10k row chunks (preserving headers)
2. Extracting FASTA sequences for left and right coordinates using bedtools
3. Adding extracted sequences as new columns (lseq, rseq)
4. Performing MFE calculation on the extracted sequences (RNAduplex)
5. Optionally generating shuffled control sequences for MFE comparisons (uShuffle) and/or flipped arm control (reversing one arm's sequence)
6. Concatenating all chunks back into a single table with one header and plotting

• Nextflow >= 25.10.2
• Docker or Singularity

Important: we assume BED-style 0-based half-open coordinates are given in the input.

nextflow run main.nf \
  --input samplesheet.tsv \
  --fasta /path/to/genome.fa \
  --fai /path/to/genome.fa.fai \
  --chunk_size <n> \
  --shuffled_mfe \
  --n_shuffles <n> \
  --flipped_arm_mfe \
  --outdir results

Samplesheet format (TSV with header):

sample_id	file_path
GM12878_rep1	/path/to/GM12878_rep1.chim.txt
GM12878_rep2	/path/to/GM12878_rep2.chim.txt

nextflow run main.nf \
  --input_pattern 'data/*.chim.txt' \
  --fasta /path/to/genome.fa \
  --outdir results \
  <...>

• --fasta: Path to reference genome FASTA file
• --fai: Path to reference genome FASTA index

• --input: Path to samplesheet (TSV format)
• --input_pattern: Glob pattern for input files (e.g., "data/*.txt")

• --outdir: Output directory (default: './results')
• --chunk_size: Number of rows per chunk (default: 10000)
• --shuffled_mfe: Enable MFE calculations for shuffled control sequences (default: false)
• --n_shuffles: Number of times the sequence is shuffled (default: 100)
• --klet_shuffles: klet for sequence shuffling (default: 2)
• --flipped_arm_mfe: Enable MFE calculations for flipped arm control sequences

Input files should be tab-delimited with the following required columns:

lchr	ll	lr	lstrand	rchr	rl	rr	rstrand	name

Optional columns:

mapq

For each sample, the pipeline produces:

• {sample_id}_mfe.tsv: Final table with added columns

Output columns include all original columns plus:

• lseq: Extracted sequence for left coordinate (strand-aware)

• rseq: Extracted sequence for right coordinate (strand-aware)

• mfe : Observed minimum free energy, MFE (kcal/mol)

• dot_bracket: Dot-bracket representation of the predicted duplex structure for the observed lseq and rseq

• mean_shuffled_mfe: Mean duplex MFE across all successfully evaluated shuffled sequence pairs

• sd_shuffled_mfe: Standard deviation of duplex MFE across all successfully evaluated shuffled sequence pairs

• delta_mfe: Observed MFE minus the mean shuffled MFE

• zscore_mfe: Standardized difference between the observed MFE and the shuffled MFE distribution

• empirical_p_lower: Empirical one-sided p-value for observing an MFE this low or lower relative to the shuffled null distribution

• n_shuffles_ok: Number of shuffled sequence pairs for which duplex MFE was successfully computed

• mfe_lseq_flipped: Duplex MFE after reversing lseq only and folding it against the original rseq

• flipped_lseq_dot_bracket: Dot-bracket structure for the duplex formed by reversed lseq and original rseq

• flipped_lseq_pair: The exact sequence pair used for that fold, formatted as reversed_lseq&original_rseq

• mfe_rseq_flipped: Duplex MFE after reversing rseq only and folding it against the original lseq

• flipped_rseq_dot_bracket: Dot-bracket structure for the duplex formed by original lseq and reversed rseq

• flipped_rseq_pair: The exact sequence pair used for that fold, formatted as original_lseq&reversed_rseq

nextflow run main.nf -profile docker --input ... --fasta ...

nextflow run main.nf -profile singularity --input ... --fasta ...

Input Tables
    ↓
SPLIT_TABLE (10k rows per chunk)
    ↓
PREPARE_BED (convert to BED format)
    ↓
EXTRACT_SEQUENCES (bedtools getfasta -s)
    ↓
ADD_SEQUENCES (add lseq, rseq columns)
    ↓
CALCULATE_MFE or CALCULATE_MFE_CONTROLS (MFE-related columns)
    ↓
CONCATENATE_TABLES (merge chunks)
    ↓
PLOT_MFE_SUMMARY (per sample)
    ↓
Final Output

# Using samplesheet
nextflow run main.nf \
  --input samplesheet.tsv \
  --fasta hg38.fa \
  --chunk_size 10000 \
  --outdir results \
  -profile docker

conda activate nfcore_tools_34

nextflow run main.nf --fasta /Volumes/lab-ulej/home/users/luscomben/users/iosubi/projects/structurome_blencowe/ref/Homo_sapiens.GRCh38.dna.primary_assembly.fa --input ./data/samplesheet.tsv --outdir results -profile docker --chunk_size 10 -resume --fai /Volumes/lab-ulej/home/users/luscomben/users/iosubi/projects/structurome_blencowe/ref/Homo_sapiens.GRCh38.dna.primary_assembly.fa.fai --shuffled_mfe --n_shuffles 5 --flipped_arm_mfe

SPLIT_TABLE → [meta, chunk_0001.txt]
              [meta, chunk_0002.txt]
              [meta, chunk_0003.txt]
                     ↓
PREPARE_BED → [meta, chunk_0001.txt, chunk_0001_left.bed, chunk_0001_right.bed]
              [meta, chunk_0002.txt, chunk_0002_left.bed, chunk_0002_right.bed]
              [meta, chunk_0003.txt, chunk_0003_left.bed, chunk_0003_right.bed]
                     ↓
EXTRACT_SEQUENCES → [meta, chunk_0001.txt, left.fa, right.fa]
                    [meta, chunk_0002.txt, left.fa, right.fa]
                    [meta, chunk_0003.txt, left.fa, right.fa]
                     ↓
ADD_SEQUENCES → [meta, chunk_0001_sequences.tsv]
                [meta, chunk_0002_sequences.tsv]
                [meta, chunk_0003_sequences.tsv]

etc.