def draw_2nd_structure(srna):
for i in range(3):
name = srna.loc[i, "name"]
sequence = srna.loc[i, "sequence"]
folding = fold_rna(srna.loc[i, "sequence"])[0]
print("##############", name, "###############")
print(sequence)
print(folding)
print()
fx_test = ">{0}\n{1}\n{2}\n".format(name, sequence, folding)
textfile = open('./resources/' + srna.loc[i, "name"] + '.fx', "w")
textfile.write(fx_test)
textfile.close()
for i in range(3):
##Print Structure
plt.figure(figsize=(20, 20))
cg = forgi.load_rna('./resources/' + srna.loc[i, "name"] + '.fx',
allow_many=False)
fvm.plot_rna(cg,
text_kwargs={"fontweight": "black"},
lighten=0.7,
backbone_kwargs={"linewidth": 3})
plt.show()
# plt.savefig(srna.loc[i, "name"]+'.png')
return srna
def visualize(prim_str, pred_string, true_string, save_imgs = False, suffix_img = ''):
corr_string = balance_op_tmp(pred_string)
print(f'pred: {pred_string}')
print(f"true: {true_string}")
print(f'corr: {corr_string}')
with open('tmp_vis_pred.txt', 'w') as f:
f.write(prim_str+ os.linesep)
f.write(corr_string+ os.linesep)
f.close
with open('tmp_vis_true.txt', 'w') as f:
f.write(prim_str+ os.linesep)
f.write(true_string+os.linesep)
f.close
plt.figure(figsize = (20, 20))
plt.title('predicted')
cg = forgi.load_rna('tmp_vis_pred.txt', allow_many=False)
fvm.plot_rna(cg, text_kwargs={"fontweight":"black"}, lighten=0.7,backbone_kwargs={"linewidth":3})
if save_imgs:
plt.savefig(f'pred_{suffix_img}.jpg')
plt.show()
plt.figure(figsize = (20, 20))
plt.title('original')
cg = forgi.load_rna('tmp_vis_true.txt', allow_many=False)
fvm.plot_rna(cg, text_kwargs={"fontweight":"black"}, lighten=0.7,backbone_kwargs={"linewidth":3})
plt.savefig(f'true_{suffix_img}.jpg')
plt.show()
def plot_sample(sample):
"""Source: https://www.kaggle.com/erelin6613/openvaccine-rna-visualization"""
struct = sample['structure']
seq = sample['sequence']
bg = fgb.BulgeGraph.from_fasta_text(f'>rna1\n{struct}\n{seq}')[0]
plt.figure(figsize=(20, 8))
fvm.plot_rna(bg)
plt.title(f"RNA Structure (id: {sample.id})")
plt.show()
def draw_structure(
): #Spits out a 2D image representing the structure of the RNA sequence
for i in lines[1::3]:
cg = str(i)
fvm.plot_rna(cg,
text_kwargs={"fontweight": "black"},
lighten=0.7,
backbone_kwargs={"linewidth": 3})
plt.show()
def structure(sequence, dot_bracket, mirna_name, index):
out_dir = index + "/mirna"
with open(out_dir + "/" + mirna_name + ".db", "w") as temp_file:
temp_file.write(sequence + '\n')
temp_file.write(dot_bracket)
cg = forgi.load_rna(out_dir + "/" + mirna_name + ".db", allow_many=False)
fvm.plot_rna(cg,
text_kwargs={"fontweight": "black"},
lighten=0.7,
backbone_kwargs={"linewidth": 3})
plt.savefig("gui/src/assets/" + mirna_name + ".png")
def plot_sample(idx = 1, sequence=None, df=train):
fig = plt.figure(figsize=(15,15))
fig.patch.set_facecolor((160/255, 177/255, 198/255))
if sequence is not None:
samp = df[df.id == sequence]
else:
samp = df[df.index == idx]
rna = []
seq = samp.loc[samp.index[0], 'sequence']
struct = samp.loc[samp.index[0], 'structure']
bg = fgb.BulgeGraph.from_fasta_text(f'>rna1\n{struct}\n{seq}')[0]
fvm.plot_rna(bg,
text_kwargs={"fontweight":"bold",
},
lighten=0.8,
backbone_kwargs={"linewidth":3})
plt.show()
Sequence = train_df[train_df['id'] == Select_id]["sequence"].values[0]
structure = train_df[train_df['id'] == Select_id]["structure"].values[0]
predicted_loop_type = train_df[train_df['id'] ==
Select_id]["predicted_loop_type"].values[0]
bg, = bulge_graph.BulgeGraph.from_fasta_text('>seq\n' + Sequence + '\n' +
structure)
print("Sequence :", Sequence)
print("Structure :", structure)
print("Predicted Loop type :", predicted_loop_type)
print("Generated Loop type :", bg.to_element_string())
# In[8]:
plt.figure(figsize=(10, 10))
fvm.plot_rna(bg,
text_kwargs={"fontweight": "black"},
lighten=0.7,
backbone_kwargs={"linewidth": 3})
plt.show()
# ## Generating Graph Matrices from the Structures
# * [Referance](https://www.kaggle.com/theoviel/generating-graph-matrices-from-the-structures)
# In[9]:
def build_matrix(couples, size):
mat = np.zeros((size, size))
for i in range(size): # neigbouring bases are linked as well
if i < size - 1:
mat[i, i + 1] = 1
predicted_loop_type = train_df[train_df['id'] ==
Select_id]["predicted_loop_type"].values[0]
print("Sequence :", Sequence)
print("structure :", structure)
print("predicted_loop_type :", predicted_loop_type)
# ## Visualize RNA-2D Structure
# In[23]:
bg, = bulge_graph.BulgeGraph.from_fasta_text('>seq\n' + Sequence + '\n' +
structure)
plt.figure(figsize=(10, 10))
fvm.plot_rna(bg,
text_kwargs={"fontweight": "black"},
lighten=0.7,
backbone_kwargs={"linewidth": 3})
plt.show()
# ## Method for Graph Representation of RNA structure
# * In 2D-Visualization it is difficult to picture which section is which part of RNA. To make it easier, we will generate a Graph Structure. The neato method can take that as input and create a nice visualization of the graph:
# In[24]:
def render_neato(s, format='png', dpi=100):
p = subprocess.Popen(
['neato', '-T', format, '-o', '/dev/stdout', '-Gdpi={}'.format(dpi)],
stdout=subprocess.PIPE,
stdin=subprocess.PIPE)