AddAdapters

Purpose and Scope

AddAdapters is specifically designed for grading the performance of adapter-trimming tools. It serves two primary functions:

1. Synthetic Data Generation: Creates test datasets by adding known adapter contamination to clean reads
2. Performance Evaluation: Analyzes trimmed reads to calculate trimming accuracy metrics

randomreads.sh ref=ref.fa out=reads.fq len=150 paired reads=100k \
    mininsert=50 maxinsert=350 fragadapter1=AGATCGGAAGAGC fragadapter2=CTGTCTCTTATAC
rename.sh in=reads.fq out=renamed.fq renamebytrim interleaved

Basic Usage

addadapters.sh in=<file> in2=<file2> out=<outfile> out2=<outfile2> adapters=<file>

Operation Modes

Add Mode (default)

Synthetically contaminates clean reads with adapter sequences at random positions. Encodes the correct trimming answer in read headers for later evaluation.

Grade Mode

Evaluates adapter trimming performance by comparing actual read lengths against the encoded correct answers.

Parameters

Examples

addadapters.sh in=clean_reads.fq out=contaminated.fq adapters=adapters.fa

addadapters.sh in=reads.fq out=contaminated.fq literal=AGATCGGAAGAGC rate=0.3

addadapters.sh in=trimmed_reads.fq grade

addadapters.sh in1=r1.fq in2=r2.fq out1=cont_r1.fq out2=cont_r2.fq adapters=adapters.fa addpaired=t

# Generate realistic paired-end data with proper adapter placement
randomreads.sh ref=genome.fa out=reads.fq len=150 paired reads=100k \
    mininsert=50 maxinsert=350 fragadapter1=AGATCGGAAGAGC fragadapter2=CTGTCTCTTATAC

# Rename reads for compatibility with AddAdapters grading
rename.sh in=reads.fq out=test_data.fq renamebytrim interleaved

# Test your adapter trimmer
your_trimmer.sh in=test_data.fq out=trimmed.fq

# Grade the trimming performance
addadapters.sh in=trimmed.fq grade

Algorithm Details

Synthetic Contamination Strategy

In add mode, AddAdapters implements a controlled contamination algorithm:

• Position Selection: Randomly selects adapter insertion points within reads using uniform distribution
• Adapter Assignment: Randomly chooses from provided adapter sequences for each contaminated read
• Sequence Replacement: Replaces bases from insertion point onward with adapter sequence
• Tail Randomization: Fills remaining positions beyond adapter with random nucleotides
• Quality-Based Errors: When adderrors=true, introduces realistic sequencing errors in adapter regions based on Phred quality scores
• Answer Key Encoding: Modifies read IDs to format "original_length_correct_length" for automated evaluation

Performance Evaluation Metrics

In grade mode, the tool calculates comprehensive trimming statistics:

• True Positives: Reads with adapters that were correctly trimmed to the right length
• False Positives: Clean reads that were incorrectly trimmed (over-trimming)
• True Negatives: Clean reads that were correctly left untrimmed
• False Negatives: Contaminated reads that were not trimmed (under-trimming)
• Adapter Retention: Percentage of adapter sequences remaining after trimming
• Base-Level Accuracy: Precise measurement of trimming position accuracy

Limitations and Considerations

Single-End Reads

Suitable for single-end read benchmarking where adapter contamination occurs through random fragmentation or read-through events.

Paired-End Limitations

For paired reads, AddAdapters has significant limitations:

• No Insert Size Awareness: Cannot place adapters at biologically correct positions
• Unrealistic Contamination Pattern: Places adapters at same relative positions rather than complementary insert-based positions
• Overlap Detection Issues: Prevents testing of overlap-based adapter trimming methods

RandomReads Advantages

RandomReads addresses these limitations by:

• Generating reads from reference sequences with specified insert size distributions
• Adding adapters only when reads extend beyond insert boundaries (read-through)
• Creating realistic overlap patterns that enable overlap-based adapter detection
• Supporting fragment size-dependent contamination patterns

Memory and Performance

• Memory Footprint: Default allocation of 200MB, suitable for most datasets
• Streaming Processing: Processes reads in batches to maintain constant memory usage
• Thread Safety: Supports multi-threaded processing for improved throughput
• I/O Efficiency: Compatible with compressed input/output formats

Output Interpretation

Add Mode Output

Contaminated reads with modified headers encoding the correct trimming position:

@150_85 original_read_header
AGATCGGAAGAGCACACGTCTGAACTCCAGTCACATGAATCTCGTATGCCGTCTTCTGCTTG...
+
IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII...

Header format: "original_length_correct_remaining_length"

Grade Mode Statistics

Total output:                        10000 reads                  1500000 bases          
Perfectly Correct (% of output):     8750 reads (87.500%)        1312500 bases (87.500%)
Incorrect (% of output):             1250 reads (12.500%)        187500 bases (12.500%)

Adapters Remaining (% of adapters):  125 reads (2.500%)          18750 bases (1.250%)
Non-Adapter Removed (% of valid):    50 reads (0.500%)           7500 bases (0.571%)

Best Practices

• Use for single-end benchmarking when testing adapter trimming algorithms
• Choose realistic adapter sequences from your sequencing platform (Illumina TruSeq, NextSeq, etc.)
• Test multiple contamination rates (10%, 30%, 50%) to evaluate trimmer robustness
• Enable error introduction (adderrors=t) for realistic contamination patterns
• For paired reads, use RandomReads workflow for biologically accurate test data
• Grade mode requires AddAdapters-generated headers - it cannot evaluate arbitrary trimmed data