From 795a11073c88ebd5598cedb211892bdd13221adf Mon Sep 17 00:00:00 2001
From: Robert Hubley <rhubley@systemsbiology.org>
Date: Tue, 10 Sep 2024 12:29:54 -0700
Subject: [PATCH]  Created a new FASTA distribution of the curated portion of
 Dfam  and placed it in:

   https://www.dfam.org/releases/current/families/Dfam-RepeatMasker.lib.gz

 This can be used as a baseline library for RepeatClassifier in TETools.
 I added this as a download in getsrc and added Dockerfile commands
 to unzip it and run makeblastdb on it.
---
 Dockerfile | 30 ++++++++++++++++++++++++------
 getsrc.sh  |  1 +
 2 files changed, 25 insertions(+), 6 deletions(-)

diff --git a/Dockerfile b/Dockerfile
index f266f47..49fe523 100644
--- a/Dockerfile
+++ b/Dockerfile
@@ -123,12 +123,12 @@ RUN cd /opt \
     && make
 
 # Configure RepeatMasker
+# With Minimal TE Library
+#   - Also with full Dfam curated RepeatMasker.lib for RepeatClassifier
 RUN cd /opt \
-    && tar -x -f src/RepeatMasker-4.1.6.tar.gz \
+    && tar -x -f src/RepeatMasker-4.1.7.tar.gz \
     && chmod a+w RepeatMasker/Libraries \
     && chmod a+w RepeatMasker/Libraries/famdb \
-    && gunzip src/dfam38_full.0.h5.gz \
-    && mv src/dfam38_full.0.h5 /opt/RepeatMasker/Libraries/famdb/dfam38_full.0.h5 \
     && cd RepeatMasker \
     && perl configure \
         -hmmer_dir=/opt/hmmer/bin \
@@ -136,10 +136,28 @@ RUN cd /opt \
         -libdir=/opt/RepeatMasker/Libraries \
         -trf_prgm=/opt/trf \
         -default_search_engine=rmblast \
-    && cd .. && rm src/RepeatMasker-4.1.6.tar.gz
+    && gunzip -c src/Dfam-RepeatMasker.lib.gz > RepeatMasker/Libraries/RepeatMasker.lib \
+    && /opt/rmblast/bin/makeblastdb -dbtype nucl -in RepeatMasker/Libraries/RepeatMasker.lib \
+    && cd .. && rm src/RepeatMasker-4.1.7.tar.gz
+
+# With Dfam root partition
+#RUN cd /opt \
+#    && tar -x -f src/RepeatMasker-4.1.6.tar.gz \
+#    && chmod a+w RepeatMasker/Libraries \
+#    && chmod a+w RepeatMasker/Libraries/famdb \
+#    && gunzip src/dfam38_full.0.h5.gz \
+#    && mv src/dfam38_full.0.h5 /opt/RepeatMasker/Libraries/famdb/dfam38_full.0.h5 \
+#    && cd RepeatMasker \
+#    && perl configure \
+#        -hmmer_dir=/opt/hmmer/bin \
+#        -rmblast_dir=/opt/rmblast/bin \
+#        -libdir=/opt/RepeatMasker/Libraries \
+#        -trf_prgm=/opt/trf \
+#        -default_search_engine=rmblast \
+#    && gunzip -c src/Dfam-RepeatMasker.lib.gz > RepeatMasker/Libraries/RepeatMasker.lib \
+#    && /opt/rmblast/bin/makeblastdb -dbtype nucl -in RepeatMasker/Libraries/RepeatMasker.lib \
+#    && cd .. && rm src/RepeatMasker-4.1.6.tar.gz
 
-## Get the RepeatMasker.lib from a full installation
-## --and run makeblastdb
 
 # Include config update
 COPY tetoolsDfamUpdate.pl /opt/RepeatMasker/tetoolsDfamUpdate.pl
diff --git a/getsrc.sh b/getsrc.sh
index edc3212..31555bc 100755
--- a/getsrc.sh
+++ b/getsrc.sh
@@ -36,6 +36,7 @@ download https://github.com/TravisWheelerLab/NINJA/archive/0.99-cluster_only.tar
 # download https://www.repeatmasker.org/coseg-0.2.3.tar.gz
 download https://github.com/rmhubley/coseg/archive/refs/tags/coseg-0.2.3.tar.gz
 download https://www.dfam.org/releases/Dfam_3.8/families/FamDB/dfam38_full.0.h5.gz
+download https://www.dfam.org/releases/Dfam_3.8/families/Dfam-RepeatMasker.lib.gz
 download http://www.repeatmasker.org/RepeatMasker/RepeatMasker-4.1.6.tar.gz
 download https://github.com/Dfam-consortium/RepeatModeler/archive/2.0.5.tar.gz RepeatModeler-2.0.5.tar.gz
 # download https://github.com/zhangrengang/TEsorter/archive/v1.4.6.tar.gz TEsorter-1.4.6.tar.gz