modify pre_model_delete strategy 2

bioinfomaticsCSU · Jun 15, 2019 · 382585e · 382585e
1 parent ef18d39
commit 382585e
Show file tree

Hide file tree

Showing 2 changed files with 4 additions and 2 deletions.
diff --git a/README.md b/README.md
@@ -77,7 +77,7 @@ The example data can be downloaded from [here](http://bioinformatics.csu.edu.cn/
 Before run deepsignal, the reads must be processed by the *re-squiggle* module of [tombo](https://github.com/nanoporetech/tombo).
 
 Note:
-- If the fast5 files are in multi-read FAST5 format, please use _multi_to_single_fast5_ command from the [ont_fast5_api package](https://github.com/nanoporetech/ont_fast5_api) to conver the fast5 files first (Ref to [issue #173](https://github.com/nanoporetech/tombo/issues/173) in  [tombo](https://github.com/nanoporetech/tombo)).
+- If the fast5 files are in multi-read FAST5 format, please use _multi_to_single_fast5_ command from the [ont_fast5_api package](https://github.com/nanoporetech/ont_fast5_api) to convert the fast5 files first (Ref to [issue #173](https://github.com/nanoporetech/tombo/issues/173) in  [tombo](https://github.com/nanoporetech/tombo)).
 - If the basecall results are saved as fastq, run the [*tombo proprecess annotate_raw_with_fastqs*](https://nanoporetech.github.io/tombo/resquiggle.html) command before *re-squiggle*.
 
 For the example data:

diff --git a/deepsignal/train_model.py b/deepsignal/train_model.py
@@ -45,6 +45,8 @@ def train(train_file, valid_file, model_dir, log_dir, kmer_len, cent_signals_len
 
     model_regex = re.compile(r"bn_" + str(kmer_len) + "\.sn_" + str(cent_signals_len) +
                              "\.epoch_\d+\.ckpt*")
+    model_prefix = "bn_" + str(kmer_len) + ".sn_" + str(cent_signals_len) + ".epoch_"
+    model_suffix = '.ckpt'
     if os.path.exists(model_dir):
         # shutil.rmtree(model_dir)
         count = 0
@@ -210,7 +212,7 @@ def train(train_file, valid_file, model_dir, log_dir, kmer_len, cent_signals_len
                         test_accu_best_ep = np.mean(test_accuracy_total)
                         if test_accu_best_ep > test_accu_best:
                             saver.save(sess, "/".join([model_dir,
-                                                       ]))
+                                                       model_prefix + str(epoch_id) + model_suffix]))
 
                     end = time.time()
                     line = "epoch: %d, iterid: %d\n train_loss: %.3f, valid_loss: %.3f, train_accuracy: %.3f, " \