MAF Counter is a multithreaded tool designed to efficiently extract and count k-mers from multiple genome alignments in MAF format. It utilizes producer-consumer threading with concurrent queues to parallelize k-mer extraction and aggregation. The tool supports canonical/reverse-complement k-mer handling and provides flexible output options, including per-genome k-mer files and a consolidated single-file format.

1. Chunk Division: The input MAF file is divided into approximately equal-sized chunks, each containing whole alignment blocks, ensuring no block spans multiple chunks.
2. Parallel Processing: Producer threads process each chunk independently, using a sliding window to extract k-mers and their counts.
3. Consumer Merging: Consumer threads merge k-mer groups into a final aggregated data structure, avoiding conflicts by partitioning based on hashed k-mer keys.
4. File Writing: Once processing is complete, output is written either as per-genome files in Jellyfish format or a single compressed file if the -s flag is used.

To compile MAF Counter, use the following command:

g++ -std=c++11 -O3 -o maf_counter maf_counter.cpp -I /path/to/google_sparse_hash -I /path/to/concurrent_queue -pthread -lrt

./maf_counter [options] <k-mer length> <MAF file> <number of threads>

-c, --complement: Aggregate k-mers with their reverse complements.
-s, --single_file_output: Write all k-mers to a single compressed file.

• Extract 15mers from input.maf using 8 producers and 8 consumers ( suitable ideally for 16 processor cores )

./maf_counter 15 input.maf 16

• The same options but aggregate each kmer with its reverse complement (writing the lexicographically first) and output the results in sinle file mode

./maf_counter -c -s 15 input.maf 16

By default, the tool generates per-genome k-mer files in the results_counter directory. Each file is in Jellyfish format, containing k-mers and their counts for the corresponding genome ID.

Example file (genome1_kmer_counts.txt):

ATCGG  1401
TTGGC  1233

If the -s option is used, a single output file is created in the format:

ATCGG genomeid1:1401, genomeid2:1200
TTGGC genomeid1:1233, genomeid3:4123

This project is licensed under the GNU GPL v3.

For any questions or support, please contact

• izg5139@psu.edu
• mpp5977@psu.edu
• kap6605@psu.edu
• ioannis.mouratidis@psu.edu