diff --git a/R/GEOSoft-helper.R b/R/GEOSoft-helper.R
index 8223de6..a665078 100644
--- a/R/GEOSoft-helper.R
+++ b/R/GEOSoft-helper.R
@@ -1,7 +1,7 @@
 get_geo_soft <- function(ids, geo_type, dest_dir, ftp_over_https, handle_opts) {
     soft_data_list <- download_and_parse_soft(
         ids = ids, geo_type = geo_type,
-        dest_dir = dest_dir, 
+        dest_dir = dest_dir,
         ftp_over_https = ftp_over_https,
         handle_opts = handle_opts,
         only_meta = FALSE
@@ -22,8 +22,8 @@ new_geo_obj <- function(id, geo_type, soft_data) {
         GDS = methods::new(
             "GEOSoft",
             meta = soft_data$meta,
-            columns = soft_data$columns,
-            datatable = soft_data$data_table,
+            columns = column_to_rownames(soft_data$columns),
+            datatable = set_rownames(soft_data$data_table),
             accession = id
         ),
         GSE = methods::new(
diff --git a/R/GSEMatrix-helper.R b/R/GSEMatrix-helper.R
index cd9581b..43556d5 100644
--- a/R/GSEMatrix-helper.R
+++ b/R/GSEMatrix-helper.R
@@ -81,20 +81,24 @@ download_and_parse_annotation <- function(annotation, assay, dest_dir, ftp_over_
     )
     gpl_data <- parse_gpl_or_gsm_soft(read_lines(gpl_file_path))
     if (nrow(gpl_data$data_table)) {
+        feature_data <- set_rownames(gpl_data$data_table)
         # NCBI GEO uses case-insensitive matching between platform
         # IDs and series ID Refs
-        feature_data <- gpl_data$data_table[
+        feature_data <- feature_data[
             data.table::chmatch(
                 tolower(rownames(assay)),
-                tolower(rownames(gpl_data$data_table))
+                tolower(rownames(feature_data))
             ), ,
             drop = FALSE
         ]
         rownames(feature_data) <- rownames(assay)
         Biobase::AnnotatedDataFrame(feature_data,
-            varMetadata = gpl_data$columns
+            varMetadata = column_to_rownames(gpl_data$columns)
         )
     } else {
-        Biobase::AnnotatedDataFrame(data.frame(row.names = rownames(assay)))
+        Biobase::AnnotatedDataFrame(
+            data.frame(row.names = rownames(assay)),
+            varMetadata = column_to_rownames(gpl_data$columns)
+        )
     }
 }
diff --git a/R/get_geo_meta.R b/R/get_geo_meta.R
index 96f7d3f..78e89c2 100644
--- a/R/get_geo_meta.R
+++ b/R/get_geo_meta.R
@@ -4,7 +4,7 @@
 #' `search_geo()` cannot get all long metadata of GEO identities.
 #'
 #' @inheritParams get_geo
-#' @return A data.frame contains metadata of all ids.
+#' @return A [data.table][data.table] contains metadata of all ids.
 #' @export
 get_geo_meta <- function(ids, dest_dir = getwd(), ftp_over_https = TRUE, handle_opts = list(connecttimeout = 60L)) {
     ids <- toupper(ids)
@@ -28,7 +28,5 @@ get_geo_meta <- function(ids, dest_dir = getwd(), ftp_over_https = TRUE, handle_
         )
         data.table::setDT(meta)
     })
-    out <- data.table::rbindlist(meta_list, use.names = TRUE, fill = TRUE)
-    data.table::setDF(out)
-    out
+    data.table::rbindlist(meta_list, use.names = TRUE, fill = TRUE)
 }
diff --git a/R/parsers.R b/R/parsers.R
index 35663bc..069c59e 100644
--- a/R/parsers.R
+++ b/R/parsers.R
@@ -8,26 +8,18 @@ parse_gse_matrix <- function(file_text, gse_sample_data = NULL) {
 
     # fetch phenoData
     if (is.null(gse_sample_data)) {
-        parse_gse_matrix_sample_characteristics(meta_data$Sample)
-        data.table::setDF(
-            meta_data$Sample,
-            rownames = as.character(meta_data$Sample[["geo_accession"]])
-        )
-        pheno_data <- Biobase::AnnotatedDataFrame(
-            data = meta_data$Sample[colnames(matrix_data), , drop = FALSE]
-        )
+        gse_sample_data <- meta_data$Sample
+        parse_gse_matrix_sample_characteristics(gse_sample_data)
     } else {
         gse_sample_data <- parse_gse_soft_sample_characteristics(
             gse_sample_data[colnames(matrix_data)]
         )
-        data.table::setDF(
-            gse_sample_data,
-            rownames = gse_sample_data[["geo_accession"]]
-        )
-        pheno_data <- Biobase::AnnotatedDataFrame(
-            data = gse_sample_data[colnames(matrix_data), , drop = FALSE]
-        )
     }
+    gse_sample_data <- gse_sample_data[colnames(matrix_data),
+        on = "geo_accession"
+    ]
+    set_rownames(gse_sample_data, "geo_accession")
+    pheno_data <- Biobase::AnnotatedDataFrame(data = gse_sample_data)
 
     # fetch experiment data
     experiment_data <- Biobase::MIAME(
@@ -39,8 +31,8 @@ parse_gse_matrix <- function(file_text, gse_sample_data = NULL) {
         url = if (!is.null(meta_data$Series$web_link)) {
             meta_data$Series$web_link
         } else if (!is.null(meta_data$Series$geo_accession)) {
-            paste0(
-                "https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=",
+            sprintf(
+                "https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=%s",
                 meta_data$Series$geo_accession
             )
         } else {
@@ -113,8 +105,8 @@ parse_gse_soft <- function(file_text, entity_type = "all", only_meta = FALSE) {
         soft_data_list[[i]] <- methods::new(
             "GEOSoft",
             meta = entity_data$meta,
-            columns = entity_data$columns,
-            datatable = entity_data$data_table,
+            columns = column_to_rownames(entity_data$columns),
+            datatable = set_rownames(entity_data$data_table),
             accession = accession
         )
         cli::cli_progress_update()
@@ -155,9 +147,6 @@ parse_gpl_or_gsm_soft <- function(file_text, only_meta = FALSE) {
                 by = names(data_table)[[1L]]
             ]
         }
-        data.table::setDF(data_table, rownames = as.character(data_table[[1L]]))
-    } else {
-        data.table::setDF(data_table)
     }
 
     # parse column data
@@ -180,26 +169,18 @@ parse_gds_soft <- function(file_text, only_meta = FALSE) {
 
     # parse data_table data
     data_table <- read_data_table(file_text[-subset_lines])
-    if (nrow(data_table)) {
-        data.table::setDF(data_table, rownames = as.character(data_table[[1L]]))
-    } else {
-        data.table::setDF(data_table)
-    }
 
     # parse column data
     column_data <- parse_columns(file_text[-subset_lines], colnames(data_table))
-    data.table::setDT(column_data, keep.rownames = "V1")
 
     subset_data <- parse_gds_subset(file_text[subset_lines])
     # Merge subset data into column data
-    column_data <- merge(
-        column_data,
-        subset_data,
+    column_data <- merge(column_data, subset_data,
         by = "V1",
         all.x = TRUE, sort = FALSE
     )
     column_data <- column_data[colnames(data_table), on = "V1"]
-    column_data <- as.data.frame(column_data[, !1L], column_data$V1)
+    # column_data <- as.data.frame(column_data[, !1L], column_data$V1)
     list(
         data_table = data_table,
         meta = meta_data,
@@ -246,7 +227,7 @@ parse_gds_subset <- function(subset_file_text) {
             "subset_description",
             "subset_type"
         )
-    ][, lapply(.SD, paste0, collapse = "; "), by = V1]
+    ][, lapply(.SD, paste0, collapse = "; "), by = "V1"]
 }
 
 #' There are four different types of line that are recognized in SOFT. The
@@ -266,7 +247,7 @@ parse_gds_subset <- function(subset_file_text) {
 # names and values; For line seperated by "=", every row represents a item. But
 # every item in `columns` data should only own a value of length one, so we
 # collapse it.
-#' @return a data.frame
+#' @return a data.table
 #' @noRd
 parse_columns <- function(file_text, target_rownames) {
     column_data <- read_column(file_text)
@@ -290,9 +271,9 @@ parse_columns <- function(file_text, target_rownames) {
         NA_character_, labelDescription,
         na = NA_character_
     )
-    data.frame(
-        labelDescription = labelDescription,
-        row.names = target_rownames
+    data.table::data.table(
+        V1 = target_rownames,
+        labelDescription = labelDescription
     )
 }
 
diff --git a/R/phenodata.R b/R/phenodata.R
index 15c9d24..52a16df 100644
--- a/R/phenodata.R
+++ b/R/phenodata.R
@@ -139,6 +139,7 @@ parse_gse_matrix_sample_characteristics <- function(sample_dt, characteristics_c
             }
         }
     }
+    sample_dt
 }
 
 #' Parse key-value pairs in GEO series soft file
@@ -174,8 +175,7 @@ parse_gsm_list <- function(gsm_list) {
         )
     }
     res <- parse_gse_soft_sample_characteristics(gsm_list)
-    data.table::setDF(res, rownames = res[["geo_accession"]])
-    res
+    set_rownames(res, "geo_accession")
 }
 
 parse_gse_soft_sample_characteristics <- function(gsm_list) {
@@ -187,7 +187,6 @@ parse_gse_soft_sample_characteristics <- function(gsm_list) {
             )
         }
         data.table::setDT(sample_meta_data)
-        sample_meta_data
     })
     sample_meta_dt <- data.table::rbindlist(
         sample_meta_list,
diff --git a/R/search_geo.R b/R/search_geo.R
index 165c066..08f5f85 100644
--- a/R/search_geo.R
+++ b/R/search_geo.R
@@ -21,7 +21,7 @@
 #' @param step the number of records to fetch from the database each time. You
 #' may choose a smaller value if failed.
 #' @param interval The time interval (seconds) between each step.
-#' @return a data.frame contains the search results
+#' @return A [data.table][data.table] contains the search results
 #' @examples
 #'   rgeo::search_geo("diabetes[ALL] AND Homo sapiens[ORGN] AND GSE[ETYP]")
 #' @export
@@ -62,8 +62,6 @@ search_geo <- function(query, step = 500L, interval = 1L) {
         tail_col,
         after = ncol(name_value_pairs)
     )
-    data.table::setDF(name_value_pairs)
-    name_value_pairs
 }
 
 # this function just processed GEO searched results returned by `entrez_fetch`
diff --git a/R/utils.R b/R/utils.R
index 59ae9a2..d3c4be5 100644
--- a/R/utils.R
+++ b/R/utils.R
@@ -7,6 +7,20 @@ return_object_or_list <- function(x, names = NULL) {
     }
 }
 
+set_rownames <- function(x, var = 1L) {
+    if (nrow(x)) {
+        data.table::setDF(x, rownames = as.character(x[[var]]))
+    } else {
+        data.table::setDF(x)
+    }
+}
+
+column_to_rownames <- function(x, var = 1L) {
+    data.table::setDF(x[, .SD, .SDcols = !var],
+        rownames = as.character(x[[var]])
+    )
+}
+
 read_lines <- function(file) {
     data.table::fread(
         file = file, sep = "", header = FALSE,
diff --git a/man/get_geo_meta.Rd b/man/get_geo_meta.Rd
index 7f19e7b..392235b 100644
--- a/man/get_geo_meta.Rd
+++ b/man/get_geo_meta.Rd
@@ -28,7 +28,7 @@ will use \url{ftp://ftp.ncbi.nlm.nih.gov/geo} directly.}
 \link[curl:handle]{handle}.}
 }
 \value{
-A data.frame contains metadata of all ids.
+A \link{data.table} contains metadata of all ids.
 }
 \description{
 This is useful to combine with \code{\link[=search_geo]{search_geo()}} and filter results since
diff --git a/man/search_geo.Rd b/man/search_geo.Rd
index 59318b7..ee80c5a 100644
--- a/man/search_geo.Rd
+++ b/man/search_geo.Rd
@@ -21,7 +21,7 @@ may choose a smaller value if failed.}
 \item{interval}{The time interval (seconds) between each step.}
 }
 \value{
-a data.frame contains the search results
+A \link{data.table} contains the search results
 }
 \description{
 This function searchs \href{https://www.ncbi.nlm.nih.gov/gds}{GDS} database,