Copyright (C) University of Tartu 2015-2017
Please cite: Pajuste F-D, Kalpinski L, Möls M, Puurand T, Lepamets M, Remm M. (2017). FastGT: an alignment-free method for calling common SNVs directly from raw sequencing reads. Scientific Reports, 7:2537.
FastGT is a fast and accurate genotype caller from sequencing data.
It uses Empirical Bayes classifier for calling genotypes from the counts of polymorphism-specific k-mers.
FastGT can detect both canonical (diploid for autosomes and female X, haploid for male X and Y) and
non-canonical (0-4 alleles) genotypes.
-
Binaries
FastGT has two binaries: gmer_counter and gmer_caller. Pre-compiled binaries are in directory 'bin'. -
Compilation
Change into subdirectory 'src' and type:
make gmer_counter
make gmer_caller
- Usage
First one has to prepare the database of specific k-mers for each allele of polymorphism of interest.
Couple of pre-made k-mer databases are available from http://bioinfo.ut.ee/FastGT/
K-mers are counted from raw reads in FASTQ file using program gmer_counter:
gmer_counter -db DATABASE FASTQ_FILE(S) > COUNTS_FILE.txt
Then the genotypes are called using program gmer_caller:
gmer_caller COUNTS_FILE.txt > GENOTYPE_FILE.txt
Genotype file can be converted to VCF format:
generate_vcf.pl GENOTYPE_FILE.txt > GENOTYPE_FILE.vcf
Additional options for gmer_counter:
-db DATABASE - SNP/KMER database file
-dbb DBBINARY - binary database file
-w FILENAME - write binary database to file
-32 - use 32-bit integers for counts (default 16-bit)
--max_kmers NUM - maximum number of kmers per node
--header - print header row
--total - print the total number of kmers per node
--unique - print the number of non-zero kmers per node
--kmers - print individual kmer counts (default if no other output)
--compile_index - Add read index to database and write it to file
--distribution NUM - print kmer distribution (up to given number)
--num_threads - number of worker threads (default 24)
--low_memory - optimize for low memory usage
-D - increase debug level
Additional options for gmer_caller:
--training_size NUM - Use NUM markers for training (default 100000)
--runs NUMBER - Perfom NUMBER runs of model training (use 0 for no training)
--num_threads NUM - Use NUM threads (min 1, max 32, default 16)
--header - Print table header
--non_canonical - Output non-canonical genotypes
--prob_cutoff - probability cutoff for calling genotype (default 0)
--alternatives - Print probabilities of all alternative genotypes
--info - Print information about individual
--no_genotypes - Print only summary information, not actual genotypes
--model TYPE - Model type (full, diploid, haploid)
--params PARAMS - Model parameters (error, p0, p1, p2, coverage, size, size2)
--coverage NUM - Average coverage of reads
-D - increase debug level