Adapt export to recent big-refactor (#443)

Co-authored-by: Michael Pieler <[email protected]> Co-authored-by: Kevin M Jablonka <[email protected]>
OpenBioML · Aug 12, 2024 · 81d3d09 · 81d3d09
1 parent 1a844a7
commit 81d3d09
Show file tree

Hide file tree

Showing 111 changed files with 605 additions and 573 deletions.
diff --git a/data/tabular/ames_mutagenicity/meta.yaml b/data/tabular/ames_mutagenicity/meta.yaml
@@ -95,13 +95,13 @@ templates:
       Assistant: {#This sounds very exciting. |This sounds very interesting. !}Should it be a special {#molecule|one!}?
       User: Yes, the molecule should {mutagenic#not &NULL}be {mutagenic__names__adjective}.
       Assistant: {#Understood|Got it|Ok!}, this {SMILES__description} is {mutagenic#not &NULL}{mutagenic__names__adjective}: {SMILES#}
-    - Is the {SMILES__description} {SMILES#} {mutagenic__names__adjective}:<EOI> {mutagenic#no&yes}
+    - Is the {SMILES__description} {SMILES#} {mutagenic__names__adjective}:<EOI>{mutagenic#no&yes}
     - |-
       Task: Please classify a molecule based on the description.
       Description: A molecule that is {mutagenic__names__adjective}.
       {#Molecule |!}{SMILES__description}: {SMILES#}
       Constraint: Even if you are {#uncertain|not sure!}, you must pick either "True" or "False" without using any {#other|additional!} words.
-      Result:<EOI> {mutagenic#False&True}
+      Result:<EOI>{mutagenic#False&True}
     - |-
       Task: Please answer the multiple choice question.
       Question: Is the molecule with the {SMILES__description} {#representation of |!}{SMILES#} {mutagenic__names__adjective}?
@@ -115,7 +115,7 @@ templates:
       Constraint: Even if you are {#uncertain|not sure!}, you must pick either {%multiple_choice_enum%2%aA1} without using any {#other|additional!} words.
       Options:
       {mutagenic%}
-      Answer:<EOI> {%multiple_choice_result}
+      Answer:<EOI>{%multiple_choice_result}
     - |-
       Task: Please answer the multiple choice question.
       Question: Which molecules are {mutagenic#not &NULL}{mutagenic__names__adjective}?
@@ -129,4 +129,4 @@ templates:
       Constraint: You must select none, one or more options from {%multiple_choice_enum%2-5%aA1} without using any {#other|additional!} words.
       Options:
       {SMILES%mutagenic%}
-      Answer:<EOI> {%multiple_choice_result}
+      Answer:<EOI>{%multiple_choice_result}
diff --git a/data/tabular/ames_mutagenicity/transform.py b/data/tabular/ames_mutagenicity/transform.py
@@ -158,15 +158,15 @@ def get_and_transform_data():
 User: Yes, the molecule should {mutagenic#not &NULL}be {mutagenic__names__adjective}.
 Assistant: {#Understood|Got it|Ok!}, this {SMILES__description} is {mutagenic#not &NULL}{mutagenic__names__adjective}: {SMILES#}""",  # noqa: E501
             # Benchmarking text templates
-            "Is the {SMILES__description} {SMILES#} {mutagenic__names__adjective}:<EOI> {mutagenic#no&yes}",  # noqa: E501 for the benchmarking setup <EOI> separates input and output
+            "Is the {SMILES__description} {SMILES#} {mutagenic__names__adjective}:<EOI>{mutagenic#no&yes}",  # noqa: E501 for the benchmarking setup <EOI>separates input and output
             """Task: Please classify a molecule based on the description.
 Description: A molecule that is {mutagenic__names__adjective}.
 {#Molecule |!}{SMILES__description}: {SMILES#}
 Constraint: Even if you are {#uncertain|not sure!}, you must pick either "True" or "False" without using any {#other|additional!} words.
-Result:<EOI> {mutagenic#False&True}""",  # noqa: E501
+Result:<EOI>{mutagenic#False&True}""",  # noqa: E501
             # noqa: E501 """Task: Please {#give me|create|generate!} a {#molecule |!}{SMILES__description} based on the {#text |!}description{# below|!}.
             # Description: A molecule that is {mutagenic__names__adjective}.
-            # Result:<EOI> {SMILES#}""",  # noqa: E501
+            # Result:<EOI>{SMILES#}""",  # noqa: E501
             """Task: Please answer the multiple choice question.
 Question: Is the molecule with the {SMILES__description} {#representation of |!}{SMILES#} {mutagenic__names__adjective}?
 Constraint: Even if you are {#uncertain|not sure!}, you must pick either {%multiple_choice_enum%2%aA1} without using any {#other|additional!} words.
@@ -178,7 +178,7 @@ def get_and_transform_data():
 Constraint: Even if you are {#uncertain|not sure!}, you must pick either {%multiple_choice_enum%2%aA1} without using any {#other|additional!} words.
 Options:
 {mutagenic%}
-Answer:<EOI> {%multiple_choice_result}""",  # noqa: E501
+Answer:<EOI>{%multiple_choice_result}""",  # noqa: E501
             """Task: Please answer the multiple choice question.
 Question: Which molecules are {mutagenic#not &NULL}{mutagenic__names__adjective}?
 Constraint: You must select none, one or more options from {%multiple_choice_enum%2-5%aA1} without using any {#other|additional!} words.
@@ -190,7 +190,7 @@ def get_and_transform_data():
 Constraint: You must select none, one or more options from {%multiple_choice_enum%2-5%aA1} without using any {#other|additional!} words.
 Options:
 {SMILES%mutagenic%}
-Answer:<EOI> {%multiple_choice_result}""",  # noqa: E501
+Answer:<EOI>{%multiple_choice_result}""",  # noqa: E501
         ],
     }
 

diff --git a/data/tabular/bioavailability_ma_et_al/meta.yaml b/data/tabular/bioavailability_ma_et_al/meta.yaml
@@ -95,13 +95,13 @@ templates:
       Assistant: {#This sounds very exciting. |This sounds very interesting. !}Should it be a special {#molecule|one!}?
       User: Yes, the molecule should have a {bioavailable#low&high} {bioavailable__names__noun}.
       Assistant: {#Understood|Got it|Ok!}, this {SMILES__description} has a {bioavailable#low&high} {bioavailable__names__noun}: {SMILES#}
-    - Is the {SMILES__description} {SMILES#} {bioavailable__names__adjective}?<EOI> {bioavailable#yes&no}
+    - Is the {SMILES__description} {SMILES#} {bioavailable__names__adjective}?<EOI>{bioavailable#yes&no}
     - |-
       Task: Please classify a molecule based on the description.
       Description: Predict if the molecule has a low or high {bioavailable__names__noun}?
       {#Molecule |!}{SMILES__description}: {SMILES#}
       Constraint: Even if you are {#uncertain|not sure!}, you must pick either "low" or "high" without using any {#other|additional!} words.
-      Result:<EOI> {bioavailable#low&high}
+      Result:<EOI>{bioavailable#low&high}
     - |-
       Task: Please answer the multiple choice question.
       Question: Has the molecule with the {SMILES__description} {#representation of |!}{SMILES#} a high {bioavailable__names__noun}?
@@ -115,7 +115,7 @@ templates:
       Constraint: Even if you are {#uncertain|not sure!}, you must pick either {%multiple_choice_enum%2%aA1} without using any {#other|additional!} words.
       Options:
       {bioavailable%}
-      Answer:<EOI> {%multiple_choice_result}
+      Answer:<EOI>{%multiple_choice_result}
     - |-
       Task: Please answer the multiple choice question.
       Question: Which molecules have a high {bioavailable__names__noun}?
@@ -129,4 +129,4 @@ templates:
       Constraint: You must select none, one or more options from {%multiple_choice_enum%2-5%aA1} without using any {#other|additional!} words.
       Options:
       {SMILES%bioavailable%}
-      Answer:<EOI> {%multiple_choice_result}
+      Answer:<EOI>{%multiple_choice_result}
diff --git a/data/tabular/bioavailability_ma_et_al/transform.py b/data/tabular/bioavailability_ma_et_al/transform.py
@@ -156,15 +156,15 @@ def get_and_transform_data():
 User: Yes, the molecule should have a {bioavailable#low&high} {bioavailable__names__noun}.
 Assistant: {#Understood|Got it|Ok!}, this {SMILES__description} has a {bioavailable#low&high} {bioavailable__names__noun}: {SMILES#}""",  # noqa: E501
             # Benchmarking text templates
-            "Is the {SMILES__description} {SMILES#} {bioavailable__names__adjective}?<EOI> {bioavailable#no&yes}",  # noqa: E501 for the benchmarking setup <EOI> separates input and output
+            "Is the {SMILES__description} {SMILES#} {bioavailable__names__adjective}?<EOI>{bioavailable#no&yes}",  # noqa: E501 for the benchmarking setup <EOI>separates input and output
             """Task: Please classify a molecule based on the description.
 Description: Predict if the molecule has a low or high {bioavailable__names__noun}?
 {#Molecule |!}{SMILES__description}: {SMILES#}
 Constraint: Even if you are {#uncertain|not sure!}, you must pick either "low" or "high" without using any {#other|additional!} words.
-Result:<EOI> {bioavailable#low&high}""",  # noqa: E501
+Result:<EOI>{bioavailable#low&high}""",  # noqa: E501
             # noqa: E501 """Task: Please {#give me|create|generate!} a {#molecule |!}{SMILES__description} based on the {#text |!}description{# below|!}.
             # Description: A molecule that has a {bioavailable#low&high} {bioavailable__names__noun}.
-            # Result:<EOI> {SMILES#}""",  # noqa: E501
+            # Result:<EOI>{SMILES#}""",  # noqa: E501
             """Task: Please answer the multiple choice question.
 Question: Has the molecule with the {SMILES__description} {#representation of |!}{SMILES#} a high {bioavailable__names__noun}?
 Constraint: Even if you are {#uncertain|not sure!}, you must pick either {%multiple_choice_enum%2%aA1} without using any {#other|additional!} words.
@@ -176,7 +176,7 @@ def get_and_transform_data():
 Constraint: Even if you are {#uncertain|not sure!}, you must pick either {%multiple_choice_enum%2%aA1} without using any {#other|additional!} words.
 Options:
 {bioavailable%}
-Answer:<EOI> {%multiple_choice_result}""",  # noqa: E501
+Answer:<EOI>{%multiple_choice_result}""",  # noqa: E501
             """Task: Please answer the multiple choice question.
 Question: Which molecules have a high {bioavailable__names__noun}?
 Constraint: You must select none, one or more options from {%multiple_choice_enum%2-5%aA1} without using any {#other|additional!} words.
@@ -188,7 +188,7 @@ def get_and_transform_data():
 Constraint: You must select none, one or more options from {%multiple_choice_enum%2-5%aA1} without using any {#other|additional!} words.
 Options:
 {SMILES%bioavailable%}
-Answer:<EOI> {%multiple_choice_result}""",  # noqa: E501
+Answer:<EOI>{%multiple_choice_result}""",  # noqa: E501
         ],
     }
 

diff --git a/data/tabular/blood_brain_barrier_martins_et_al/meta.yaml b/data/tabular/blood_brain_barrier_martins_et_al/meta.yaml
@@ -118,13 +118,13 @@ templates:
       Assistant: {#This sounds very exciting. |This sounds very interesting. !}Should it be a special {#molecule|one!}?
       User: Yes, the molecule should {penetrate_BBB#not &NULL}be {penetrate_BBB__names__adjective}.
       Assistant: Got it, this {SMILES__description} is {penetrate_BBB#not &NULL}{penetrate_BBB__names__adjective}: {SMILES#}
-    - Is the {SMILES__description} {SMILES#} {penetrate_BBB__names__adjective}:<EOI> {penetrate_BBB#no&yes}
+    - Is the {SMILES__description} {SMILES#} {penetrate_BBB__names__adjective}:<EOI>{penetrate_BBB#no&yes}
     - |-
       Task: Please classify a molecule based on the description.
       Description: A molecule that is {penetrate_BBB__names__adjective}.
       {#Molecule |!}{SMILES__description}: {SMILES#}
       Constraint: Even if you are {#uncertain|not sure!}, you must pick either "True" or "False" without using any {#other|additional!} words.
-      Result:<EOI> {penetrate_BBB#False&True}
+      Result:<EOI>{penetrate_BBB#False&True}
     - |-
       Task: Please answer the multiple choice question.
       Question: Is the molecule with the {SMILES__description} of {SMILES#} {penetrate_BBB__names__adjective}?
@@ -145,11 +145,11 @@ templates:
       Constraint: Even if you are {#uncertain|not sure!}, you must pick either {%multiple_choice_enum%2%aA1} without using any {#other|additional!} words.
       Options:
       {penetrate_BBB%}
-      Answer:<EOI> {%multiple_choice_result}
+      Answer:<EOI>{%multiple_choice_result}
     - |-
       Task: Please answer the multiple choice question.
       Question: Which molecules are {penetrate_BBB#not &NULL}{penetrate_BBB__names__adjective}?
       Constraint: You must select none, one or more options from {%multiple_choice_enum%2-5%aA1} without using any {#other|additional!} words.
       Options:
       {SMILES%penetrate_BBB%}
-      Answer:<EOI> {%multiple_choice_result}
+      Answer:<EOI>{%multiple_choice_result}
diff --git a/data/tabular/blood_brain_barrier_martins_et_al/transform.py b/data/tabular/blood_brain_barrier_martins_et_al/transform.py
@@ -178,16 +178,16 @@ def get_and_transform_data():
 User: Yes, the molecule should {penetrate_BBB#not &NULL}be {penetrate_BBB__names__adjective}.
 Assistant: Got it, this {SMILES__description} is {penetrate_BBB#not &NULL}{penetrate_BBB__names__adjective}: {SMILES#}""",  # noqa: E501
             # Benchmarking text templates
-            "Is the {SMILES__description} {SMILES#} {penetrate_BBB__names__adjective}:<EOI> {penetrate_BBB#no&yes}",  # noqa: E501 for the benchmarking setup <EOI> separates input and output
+            "Is the {SMILES__description} {SMILES#} {penetrate_BBB__names__adjective}:<EOI>{penetrate_BBB#no&yes}",  # noqa: E501 for the benchmarking setup <EOI>separates input and output
             # todo: check if we go for multiple choice only and remove the benchmarking template above and below
             """Task: Please classify a molecule based on the description.
 Description: A molecule that is {penetrate_BBB__names__adjective}.
 {#Molecule |!}{SMILES__description}: {SMILES#}
 Constraint: Even if you are {#uncertain|not sure!}, you must pick either "True" or "False" without using any {#other|additional!} words.
-Result:<EOI> {penetrate_BBB#False&True}""",  # noqa: E501
+Result:<EOI>{penetrate_BBB#False&True}""",  # noqa: E501
             # noqa: E501 """Task: Please {#give me|create|generate!} a molecule {SMILES__description} based on the {#text |!}description{# below|!}.
             # Description: A molecule that is {penetrate_BBB__names__adjective}.
-            # Result:<EOI> {SMILES#}""",  # noqa: E501
+            # Result:<EOI>{SMILES#}""",  # noqa: E501
             """Task: Please answer the multiple choice question.
 Question: Is the molecule with the {SMILES__description} of {SMILES#} {penetrate_BBB__names__adjective}?
 Constraint: Even if you are {#uncertain|not sure!}, you must pick either {%multiple_choice_enum%2%aA1} without using any {#other|additional!} words.
@@ -205,13 +205,13 @@ def get_and_transform_data():
 Constraint: Even if you are {#uncertain|not sure!}, you must pick either {%multiple_choice_enum%2%aA1} without using any {#other|additional!} words.
 Options:
 {penetrate_BBB%}
-Answer:<EOI> {%multiple_choice_result}""",  # noqa: E501
+Answer:<EOI>{%multiple_choice_result}""",  # noqa: E501
             """Task: Please answer the multiple choice question.
 Question: Which molecules are {penetrate_BBB#not &NULL}{penetrate_BBB__names__adjective}?
 Constraint: You must select none, one or more options from {%multiple_choice_enum%2-5%aA1} without using any {#other|additional!} words.
 Options:
 {SMILES%penetrate_BBB%}
-Answer:<EOI> {%multiple_choice_result}""",  # noqa: E501
+Answer:<EOI>{%multiple_choice_result}""",  # noqa: E501
         ],
     }
 

diff --git a/data/tabular/carcinogens/meta.yaml b/data/tabular/carcinogens/meta.yaml
@@ -107,13 +107,13 @@ templates:
       Assistant: {#This sounds very exciting. |This sounds very interesting. !}Should it be a special {#molecule|one!}?
       User: Yes, the molecule should {carcinogen#not &NULL}be {carcinogen__names__adjective}.
       Assistant: {#Understood|Got it|Ok!}, this {SMILES__description} is {carcinogen#not &NULL}{carcinogen__names__adjective}: {SMILES#}
-    - Is the {SMILES__description} {SMILES#} {carcinogen__names__adjective}:<EOI> {carcinogen#no&yes}
+    - Is the {SMILES__description} {SMILES#} {carcinogen__names__adjective}:<EOI>{carcinogen#no&yes}
     - |-
       Task: Please classify a molecule based on the description.
       Description: A molecule that is {carcinogen__names__adjective}.
       {#Molecule |!}{SMILES__description}: {SMILES#}
       Constraint: Even if you are {#uncertain|not sure!}, you must pick either "True" or "False" without using any {#other|additional!} words.
-      Result:<EOI> {carcinogen#False&True}
+      Result:<EOI>{carcinogen#False&True}
     - |-
       Task: Please answer the multiple choice question.
       Question: Is the molecule with the {SMILES__description} {#representation of |!}{SMILES#} {carcinogen__names__adjective}?
@@ -127,7 +127,7 @@ templates:
       Constraint: Even if you are {#uncertain|not sure!}, you must pick either {%multiple_choice_enum%2%aA1} without using any {#other|additional!} words.
       Options:
       {carcinogen%}
-      Answer:<EOI> {%multiple_choice_result}
+      Answer:<EOI>{%multiple_choice_result}
     - |-
       Task: Please answer the multiple choice question.
       Question: Which molecules are {carcinogen#not &NULL}{carcinogen__names__adjective}?
@@ -141,4 +141,4 @@ templates:
       Constraint: You must select none, one or more options from {%multiple_choice_enum%2-5%aA1} without using any {#other|additional!} words.
       Options:
       {SMILES%carcinogen%}
-      Answer:<EOI> {%multiple_choice_result}
+      Answer:<EOI>{%multiple_choice_result}
diff --git a/data/tabular/carcinogens/transform.py b/data/tabular/carcinogens/transform.py
@@ -173,15 +173,15 @@ def get_and_transform_data():
 User: Yes, the molecule should {carcinogen#not &NULL}be {carcinogen__names__adjective}.
 Assistant: {#Understood|Got it|Ok!}, this {SMILES__description} is {carcinogen#not &NULL}{carcinogen__names__adjective}: {SMILES#}""",  # noqa: E501
             # Benchmarking text templates
-            "Is the {SMILES__description} {SMILES#} {carcinogen__names__adjective}:<EOI> {carcinogen#no&yes}",  # noqa: E501 for the benchmarking setup <EOI> separates input and output
+            "Is the {SMILES__description} {SMILES#} {carcinogen__names__adjective}:<EOI>{carcinogen#no&yes}",  # noqa: E501 for the benchmarking setup <EOI>separates input and output
             """Task: Please classify a molecule based on the description.
 Description: A molecule that is {carcinogen__names__adjective}.
 {#Molecule |!}{SMILES__description}: {SMILES#}
 Constraint: Even if you are {#uncertain|not sure!}, you must pick either "True" or "False" without using any {#other|additional!} words.
-Result:<EOI> {carcinogen#False&True}""",  # noqa: E501
+Result:<EOI>{carcinogen#False&True}""",  # noqa: E501
             # noqa: E501 """Task: Please {#give me|create|generate!} a {#molecule |!}{SMILES__description} based on the {#text |!}description{# below|!}.
             # Description: A molecule that is {carcinogen__names__adjective}.
-            # Result:<EOI> {SMILES#}""",  # noqa: E501
+            # Result:<EOI>{SMILES#}""",  # noqa: E501
             """Task: Please answer the multiple choice question.
 Question: Is the molecule with the {SMILES__description} {#representation of |!}{SMILES#} {carcinogen__names__adjective}?
 Constraint: Even if you are {#uncertain|not sure!}, you must pick either {%multiple_choice_enum%2%aA1} without using any {#other|additional!} words.
@@ -193,7 +193,7 @@ def get_and_transform_data():
 Constraint: Even if you are {#uncertain|not sure!}, you must pick either {%multiple_choice_enum%2%aA1} without using any {#other|additional!} words.
 Options:
 {carcinogen%}
-Answer:<EOI> {%multiple_choice_result}""",  # noqa: E501
+Answer:<EOI>{%multiple_choice_result}""",  # noqa: E501
             """Task: Please answer the multiple choice question.
 Question: Which molecules are {carcinogen#not &NULL}{carcinogen__names__adjective}?
 Constraint: You must select none, one or more options from {%multiple_choice_enum%2-5%aA1} without using any {#other|additional!} words.
@@ -205,7 +205,7 @@ def get_and_transform_data():
 Constraint: You must select none, one or more options from {%multiple_choice_enum%2-5%aA1} without using any {#other|additional!} words.
 Options:
 {SMILES%carcinogen%}
-Answer:<EOI> {%multiple_choice_result}""",  # noqa: E501
+Answer:<EOI>{%multiple_choice_result}""",  # noqa: E501
         ],
     }