update descriptions and channels

ktmeaton · May 6, 2020 · d1567af · d1567af
1 parent d713d24
commit d1567af
Show file tree

Hide file tree

Showing 2 changed files with 20 additions and 9 deletions.
diff --git a/docs/process/process_data_download.rst b/docs/process/process_data_download.rst
@@ -4,37 +4,46 @@ Data Download
 SQLite Import
 ------------------
 
-Import assembly FTP url from database, also retrieve file names for web get.
+Import assembly FTP url from database, retrieve file names for web get, prepare TSV input of SRA metadata for EAGER pipeline.
 
 ========================================= =========================== ===========================
 Input                                     Type                        Description
 ========================================= =========================== ===========================
-ch_sqlite                                 sqlite                      NCBImeta SQLite database from process ncbimeta_db_update or params.sqlite
+ch_sqlite                                 sqlite                      NCBImeta SQLite database from params.sqlite or process :ref:`ncbimeta_db_update<NCBImeta DB Update>`
 ========================================= =========================== ===========================
 
 ========================================= =========================== ===========================
 Output                                    Type                        Description
 ========================================= =========================== ===========================
-ch_assembly_for_download_ftp              url                         FTP url for process assembly_download.
+ch_assembly_for_download_ftp              text                        FTP url for process assembly_download :ref:`assembly_download<Assembly Download>`.
+ch_sra_tsv_for_eager                      tsv                         TSV metadata input for process eager.
 ========================================= =========================== ===========================
 
 ========================================= =========================== ===========================
 Publish                                    Type                        Description
 ========================================= =========================== ===========================
 file_assembly_for_download_ftp            text                        List of FTP urls for genomic assembly download.
+${params.eager_tsv}                       tsv                         TSV metadata input for EAGER pipeline.
 ========================================= =========================== ===========================
 
 **Shell script**::
 
-      sqlite3 ${sqlite} ${params.sqlite_select_command} | grep . | head -n ${params.max_datasets} | sed 's/ /\\n/g' | while read line;
+      # Select the Genbank Assemblies
+      sqlite3 ${sqlite} ${params.sqlite_select_command_asm} | grep . | head -n ${params.max_datasets} | sed -E -e 's/ |;/\\n/g' | while read line;
       do
         if [[ ! -z \$line ]]; then
-          asm_url=\$line;
-          asm_fasta=`echo \$line | cut -d "/" -f 10 | awk -v suffix=${params.genbank_asm_gz_suffix} '{print \$0 suffix}'`;
-          asm_ftp=\${asm_url}/\${asm_fasta};
+          asm_ftp=`echo \$line | \
+              awk -F "/" -v suffix=${params.genbank_assembly_gz_suffix} '{print \$0 FS \$NF suffix}'`;
           echo \$asm_ftp >> ${params.file_assembly_for_download_ftp}
         fi;
       done;
+      # Extract SRA Metadata for EAGER tsv
+      ${params.scriptdir}/sqlite_EAGER_tsv.py \
+        --database ${sqlite} \
+        --query ${params.sqlite_select_command_sra} \
+        --organism ${params.eager_organism} \
+        --max-datasets ${params.max_datasets} \
+        --output ${params.eager_tsv}
 
 ------------
 

diff --git a/pipeline.nf b/pipeline.nf
@@ -228,16 +228,18 @@ if( (params.sqlite || ( params.ncbimeta_update && params.ncbimeta_annot) ) && !p
 
   process sqlite_import{
     /*
-    Import assembly FTP url from database, also retrieve file names for web get.
+    Import assembly FTP url from database, retrieve file names for web get, prepare TSV input of SRA metadata for EAGER pipeline.
 
     Input:
     ch_sqlite (sqlite): NCBImeta SQLite database from process ncbimeta_db_update or params.sqlite
 
     Output:
-    ch_assembly_for_download_ftp (url): FTP url for process assembly_download.
+    ch_assembly_for_download_ftp (text): FTP url for process assembly_download.
+    ch_sra_tsv_for_eager (tsv): TSV metadata input for process eager.
 
     Publish:
     file_assembly_for_download_ftp (text): List of FTP urls for genomic assembly download.
+    ${params.eager_tsv} (tsv): TSV metadata input for EAGER pipeline.
     */
     // Other variables and config
     tag "$sqlite"