spASM

Tool for finding allele-specific methylation with the epiBED format.

Install

spASM is written in Rust. If Rust is not installed, you can find instructions here.

For now, spASM can be installed by cloning the repository and then built with cargo (installed when installing Rust).

$ git clone [email protected]:jamorrison/spASM.git
$ cd spASM
$ cargo build --release

By default, Rust builds in debug mode, so you will have to include --release to build in release mode with the optimizations that are included therein. The default path the binary will be built at is (spasm top directory)/target/release/spasm.

When development slows, static binaries will be made available on the GitHub releases page.

Usage

spasm [options] <GENOME> <PATH>

Required Arguments

Name	Description
GENOME	Indexed FASTA (samtools faidx) file of your genome
PATH	Path to epiBED file (if running with -g, then it must be bgzip'd + tabix'd)

Optional Arguments

Short Option (-)	Long Option (--)	Description	Default
g	region	region to extract (chr:start-end or chr)	all
n	no-mate-merging	do not merge mate reads together into a single DNA fragment	mate reads are merged together
c	fdr	type of false discovery rate correction to perform possibilities: BH (Benjamini-Hochberg), BY (Benjamini-Yekutieli), Bonferroni, Hochberg, Holm, No (do not apply false discovery correction)	BH
p	pcutoff	p-value significance cutoff	0.05
o	output	output file name, compression level based on file name	stdout
O	candidate	write only candidate locations, FDR-corrected p-value based on all locations probed	all locations printed
N	no-ambiguous	write only locations with no ambiguous SNPs	all SNPs are written
b	biscuit	write in BISCUIT ASM output format	output written in BEDPE format
v	verbose	verbosity level (0: ERRORS ONLY, 1: WARNINGS + ERRORS, 2+: ALL)	1
h	help	Print help
V	version	Print version

Merging Mates in Paired-End Data

In paired-end sequencing, the two read mates come from the same DNA fragment; therefore, they represent the same "epi-haplotype." To recover this correlation, mate reads are merged into a single fragment. On a qualitative level, spASM will take the unique portions of reads 1 and 2, plus the read 1 portion of any locations that overlap between the two reads. With respect to those overlapping regions, it should be noted:

• spASM will use the data as it comes from biscuit epiread. By default, biscuit epiread will filter overlapping bases (including CpGs and SNPs) from read 2. This will then be passed to spASM, which will see these as filtered bases and won't include them in any calculations.
• The only way to double count information from overlapping portions of reads 1 and 2 would be to include the -d in biscuit epiread.

Output

spASM can produce output in one of two formats. The default method is a BEDPE-compliant format with the following columns:

1. SNP chromosome
2. SNP start (0-based)
3. SNP end (1-based, non-inclusive)
4. CpG chromosome
5. CpG start (0-based)
6. CpG end (1-based, non-inclusive)
7. "candidate" (if < p-value cutoff) or "non_candidate" (if >= p-value cutoff)
8. p-value after applied false discovery correction
9. SNP₁ (top row in contingency table)
10. SNP₂ (bottom row in contigency table)
11. CpG₁ (left column in contingency table)
12. CpG₂ (right column in contigency table)
13. SNP₁-CpG₁ value in contingency table
14. SNP₁-CpG₂ value in contingency table
15. SNP₂-CpG₁ value in contingency table
16. SNP₂-CpG₂ value in contingency table

spASM can also produce output in the format generated by biscuit asm:

1. Chromosome
2. SNP position (0-based)
3. CpG position (0-based)
4. SNP₁ / SNP₂
5. CpG₁ / CpG₂
6. SNP₁-CpG₁ value in contingency table
7. SNP₁-CpG₂ value in contingency table
8. SNP₂-CpG₁ value in contingency table
9. SNP₂-CpG₂ value in contingency table
10. p-value after applied false discovery correction
11. . (unused column included for consistency with biscuit format)