Skip to contents

Create a Modifications DuckDB from Parquet CH3 files (with optional sample naming)

Source: R/make_mod_db.R
make_mod_db.Rd

Build a DuckDB database containing filtered modification call data from one or more .ch3 parquet files. Inputs may be individual files, directories (expanded to all *.ch3 files), or a **named character vector** where names are used as sample_names in the output.

Usage

make_mod_db(
  ch3_files,
  db_name,
  chrom = NULL,
  min_read_length = 50,
  min_call_prob = 0.9,
  min_base_qual = 10,
  flag = NULL
)

Arguments

ch3_files

Character vector of CH3 parquet file paths and/or directories. May be a named vector to assign sample_names explicitly; any entry of the form NAME=PATH is also accepted. Directories are scanned (non-recursively) for *.ch3 files. Must not be empty.

db_name

Path (without or with .mod.db extension) for the DuckDB database to be created; .mod.db is appended if missing.

chrom

Optional chromosome filter. Either a single string (e.g., "chr1") or a character vector (e.g., c("chr1","chr2","chrX")). If NULL, all chromosomes are included.

min_read_length

Minimum read length to keep (default 50).

min_call_prob

Minimum call probability to keep (default 0.9).

min_base_qual

Minimum base quality to keep (default 10).

flag

Optional numeric flag value to require; if NULL, no flag filter.

Value

(Invisibly) a list of class "mod_db" with elements:

  • db_file: path to the created DuckDB file,

  • current_table: NULL (set by downstream functions),

  • con: connection is closed by cleanup and set to "none".

The database contains at least the calls table.

Details

What it does

  • Expands ch3_files (handling directories and named entries) into a mapping of source files and optional sample_names.

  • Configures DuckDB pragmas for temp directory, thread count (all-but-one core), and a memory limit (~50

  • Drops any existing tables in the target DB.

  • Reads all input .ch3 parquet files in a single pass and creates a table calls with columns: sample_name, chrom, start, end, read_position, call_code, read_length, call_prob, base_qual, flag. When names are not given for inputs, sample_name defaults to the file stem.

  • Applies pushdown filters based on chrom, min_read_length, min_call_prob, min_base_qual, and flag.

Side effects and performance

  • Creates (or overwrites) a DuckDB file at db_name.

  • Uses a temp directory for DuckDB spills under tempdir().

  • A temporary in-memory table file_map may be created for input mapping.

Input forms

  • Files: c("a.ch3", "b.ch3")

  • Directories: c("dir_of_ch3s/") (expands to all *.ch3)

  • Named files/dirs: c(SampleA = "a.ch3", SampleB = "dir/") — names become sample_name. If a name is not provided, the filename stem (without .ch3) is used.

See also

summarize_mod_positions, summarize_mod_regions, summarize_mod_windows, get_mod_dbinfo, get_mod_tableinfo, calc_mod_diff

Examples

if (FALSE) { # \dontrun{
# 1) Directory of CH3 files (non-recursive scan for *.ch3)
make_mod_db(ch3_files = "path/to/ch3_dir",
            db_name   = "my_db")

# 2) Explicit files (auto-sample names from stems)
make_mod_db(ch3_files = c("A.ch3", "B.ch3"),
            db_name   = "two_samples.mod.db",
            min_read_length = 100, min_base_qual = 10)

# 3) Named inputs (sample_name set from names)
make_mod_db(
  ch3_files = c(
    Sample1      = "../CH3/Sample1.ch3",
    Sample2  = "../CH3/Sample2.ch3"
  ),
  db_name = "My_DB",
  min_base_qual = 10,
  min_read_length = 100
)

# 4) Filter to specific chromosomes
make_mod_db(
  ch3_files = c(S1 = "A.mod", S2 = "B.mod"),
  db_name   = "chr1_chrX_only",
  chrom     = c("chr1","chrX")
)
} # }