From 7fde40de8a54e6bbcfc7f6490c2b86fe57f4dbb4 Mon Sep 17 00:00:00 2001
From: chenkenbio <chenkenbio@gmail.com>
Date: Fri, 2 Jun 2023 17:01:54 +0800
Subject: [PATCH] add download.sh

Former-commit-id: 342c17b0c226d40d46c4142ad915015c984678c0
---
 README.md   | 10 +++-------
 download.sh | 23 +++++++++++++++++++++++
 2 files changed, 26 insertions(+), 7 deletions(-)
 create mode 100644 download.sh

diff --git a/README.md b/README.md
index 2e44559..0a3b0df 100644
--- a/README.md
+++ b/README.md
@@ -22,11 +22,7 @@ See [official guide](https://huggingface.co/docs/transformers/model_doc/bert) fo
 
 **Download SpliceBERT**
 
-- [SpliceBERT.1024nt.tar.gz](https://github.com/biomed-AI/SpliceBERT/releases/download/v0.1/SpliceBERT.1024nt.tar.gz)
-- [SpliceBERT.510nt.tar.gz](https://github.com/biomed-AI/SpliceBERT/releases/download/v0.1/SpliceBERT.510nt.tar.gz)
-- [SpliceBERT-human.510nt.tar.gz](https://github.com/biomed-AI/SpliceBERT/releases/download/v0.1/SpliceBERT-human.510nt.tar.gz)
-
- 	The model weights are also available at [zenodo](https://doi.org/10.5281/zenodo.7995778).
+The weights of SpliceBERT can be downloaded from [zenodo](https://doi.org/10.5281/zenodo.7995778): https://zenodo.org/record/7995778/files/models.tar.gz?download=1
 
 **System requirements**  
 
@@ -69,8 +65,8 @@ model = AutoModelForSequenceClassification.from_pretrained(SPLICEBERT_PATH, num_
 
 ## Reproduce the analysis in manuscript  
 
-Before running the codes, run `bash setup.sh` in the `./examples` folder to compile the codes written in cython (`cython` is required).  
-Then, run `bash download.sh` to fetch the data used in the analysis.
+Before running the codes, run `bash download.sh` to fetch the data used in the analysis.
+Then, run `bash setup.sh` in the `./examples` folder to compile the codes written in cython (`cython` is required).  
 
 The codes for analyzing SpliceBERT are available in [examples](./examples):  
 - [evolutionary conservation analysis](./examples/00-conservation) (related to Figure 1)  
diff --git a/download.sh b/download.sh
new file mode 100644
index 0000000..417421e
--- /dev/null
+++ b/download.sh
@@ -0,0 +1,23 @@
+#!/bin/bash
+
+
+echo "Downloading the data to ./examples/ ..."
+wget -c -O ./examples/data.tar.gz https://zenodo.org/record/7995778/files/data.tar.gz?download=1 && cd examples && tar -xzvf data.tar.gz && cd .. && echo "Done"
+
+echo "Downloading the model weights ..."
+wget -c -O models.tar.gz https://zenodo.org/record/7995778/files/models.tar.gz?download=1 && tar -xzvf models.tar.gz && echo "Done"
+
+
+## check dnabert
+echo "Preparing the DNABERT weights ..."
+test -d ./models/dnabert || mkdir -p ./models/dnabert
+cd ./models/dnabert
+for k in 3 4 5 6; do
+    if [ ! -e "${k}-new-12w-0" ]; then
+        if [ -e "${k}-new-12w-0.zip" ]; then
+            unzip "${k}-new-12w-0.zip" && echo "unzip: ${k}-new-12w-0.zip -> ${k}-new-12w-0"
+        else
+            echo "NOTE: Users should manually download the weights of DNABERT${k} from https://github.com/jerryji1993/DNABERT and decompress it to ./models/dnabert/"
+        fi
+    fi
+done