R/sqlite_view_cdhits.R

In thierrycnam/igfuns: Analyzes of Ig sequencing data

#' sqlite_view_cdhits
#'
#' Create a view containing coverage information based on cdhits table.
#'
#' @param cdhit_dbtype dbtype
#' @param cdhit_fraction fraction
#' @param view_name viewname
#'
#' @return
#' @export
#'
#' @examples
#' sqlite_view_cdhits(cdhit_dbtype = "nt", cdhit_fraction = 975)
sqlite_view_cdhits <- function(view_name = "cds_100",
                               cdhit_dbtype = "nt",
                               cdhit_fraction=100) {
  sql_query <-sprintf('
    CREATE VIEW IF NOT EXISTS %s
    AS
    SELECT cds.*, cds_count.n
    FROM cdhits cds
    JOIN
    (SELECT clstr_id, COUNT(*) AS n
    FROM cdhits cds
    GROUP BY clstr_id) cds_count
    ON cds.clstr_id = cds_count.clstr_id
    WHERE cds.is_clstr_ref = 1
    AND cds.fraction = %s
    AND cds.dbtype = "%s";', view_name, cdhit_fraction, cdhit_dbtype)
  sql_query
}