Misc changes

README.md

@@ -131,6 +131,8 @@ nwr ardb --genbank
 
 nwr common "Escherichia coli" 4932 Drosophila_melanogaster 9606 "Mus musculus"
 
+# rm ~/.nwr/*.dmp
+
 ```
 
 ### Development
@@ -158,9 +160,17 @@ cargo run --bin nwr template tests/assembly/Trichoderma.assembly.tsv --ass -o st
 ### seqdb
 
 ```shell
-export SEQ_DIR="$HOME/data/Bacteria/Protein/Zymomonas_mobilis"
+export SEQ_DIR="$HOME/data/Bacteria/Protein/Pseudomonas_fragi"
 #export SEQ_DIR="$HOME/data/Bacteria/Protein/Pseudomonas_aeruginosa"
 
+cargo run --bin nwr seqdb -d ${SEQ_DIR} --init --strain
+
+cargo run --bin nwr seqdb -d ${SEQ_DIR} \
+    --size <(
+        hnsm size ${SEQ_DIR}/pro.fa.gz
+    ) \
+    --clust
+
 cat "${SEQ_DIR}"/strains.tsv |
     parallel --colsep '\t' --no-run-if-empty --linebuffer -k -j 1 '
         if [[ ! -d "$HOME/data/Bacteria/ASSEMBLY/{2}/{1}" ]]; then
@@ -183,14 +193,6 @@ cat "${SEQ_DIR}"/strains.tsv |
     ' \
     > "${SEQ_DIR}"/detail.tsv
 
-cargo run --bin nwr seqdb -d ${SEQ_DIR} --init --strain
-
-cargo run --bin nwr seqdb -d ${SEQ_DIR} \
-    --size <(
-        hnsm size ${SEQ_DIR}/pro.fa.gz
-    ) \
-    --clust
-
 cargo run --bin nwr seqdb -d ${SEQ_DIR} \
     --anno <(
         tsv-select -f 1,3 "${SEQ_DIR}"/detail.tsv | tsv-uniq

src/cmd_nwr/seqdb.rs

@@ -21,6 +21,8 @@ the protein sequences are identical or highly similar
 
 * If `--strain` is called without specifying a path, it will load the default file under `--dir`
 
+* `--rep` requires two arguemnts, `--rep f1 file`
+
 * The DDL
 
 {}
@@ -70,6 +72,12 @@ the protein sequences are identical or highly similar
                 .num_args(0..=1)
                 .help("Load asmseq.tsv"),
         )
+        .arg(
+            Arg::new("rep")
+                .long("rep")
+                .num_args(2)
+                .help("Load features into rep"),
+        )
 }
 
 static DDL_SEQ: &str = r###"