New Comments in Tutorial

get_ipython().getoutput("curl -Lo conda_installer.py https://raw.githubusercontent.com/deepchem/deepchem/master/scripts/colab_install.py")

import conda_installer

conda_installer.install()

get_ipython().getoutput("/root/miniconda/bin/conda info -e")

get_ipython().getoutput("pip install --pre deepchem")

get_ipython().getoutput("pip install propy3 ")

#imports 

import numpy as np 

import pandas as pd

import matplotlib.pyplot as plt

data =pd.read_csv("./all_measurementsProtherDB.tsv",delimiter="\t")

data.shape

df = data[ (data['MEASURE'] == 'DSC') & (data['MUTATION'] == 'wild-type')]

df.shape

no_mutants_dataset = df.drop_duplicates(['PDB_wild'])

no_mutants_dataset.shape

df2 = data[ (data['MEASURE'] == 'DSC')]

df2.shape

no_wild_type_repeated  =df2.drop_duplicates(['MUTATION','PDB_wild'],keep='first')

no_wild_type_repeated.sort_values('UniProt_ID')[['UniProt_ID','PDB_wild','PROTEIN']]

Mutation_list = no_wild_type_repeated[['UniProt_ID','MUTATION']].set_index('UniProt_ID')

Mutation_list

import requests, sys

def Seq_From_AccNum(num):

    '''

    perform a http  request to the EBI-API to obtain the sequence based on the Uniprot Accession number

    return a string 

    '''

    requestURL = "https://www.ebi.ac.uk/proteins/api/features/{}".format(num)

    r = requests.get(requestURL, headers={ "Accept" : "application/json"})

    if not r.ok:

        print("Failure in Uniprot request")

        return None

        r.raise_for_status()

        sys.exit()

    responseBody = r.json()

    return responseBody['sequence']

import re

def MutateSeq(seq,Mutant):

    '''

    mutate a sequence based on a string (Mutant) that has the notation : 

    A###B where A is the wildtype aminoacid ### the position and B the mutation

    '''

    aalist = re.findall('([A-Z])([0-9]+)([A-Z])', Mutant)

    #(len(aalist)==1):

    newseq=seq

    listseq=list(newseq)

    for aas in aalist:

        wildAA = aas[0]

        pos = int(aas[1]) -1

        if(pos >= len(listseq)):

            print("Mutation not in the range of the protein")

            return None

        MutAA = aas[-1]

        if(listseq[pos]==wildAA):

            listseq[pos]=MutAA

        else:

            print("WildType AA does not match")

            return None

    return("".join(listseq))

Mutation_list.sort_values('UniProt_ID',inplace=True)

Mutation_list

import time 

t0 = time.time()

Sequences = {}

fail  = {}

AccNum = []

count = 0 

print("Perfoming data curation : ")

for accnumber, row in Mutation_list.iterrows():

    #if(count == 100):

    #    break

    #count += 1

    mutation = row['MUTATION'].split("(")[0].strip()

    print("{} - {}".format(accnumber,mutation), end =" ")

    name ="{}-{}".format(accnumber,mutation)

    if(accnumber=='-'):

        fail[accnumber] = [mutation]

        continue

    if accnumber not in AccNum:

        AccNum.append(accnumber)

        seq = Seq_From_AccNum(accnumber)

        if(seq == None):

            continue

    if(mutation =='wild-type'):

        name ="{}-{}".format(accnumber,"WT")

        Sequences[name]=seq

    else:

        mutseq = MutateSeq(seq,mutation)

        if(mutseq==None):

            if(accnumber not in fail ):

                fail[accnumber] = [mutation]

            else:

                fail[accnumber].append(mutation)

        Sequences[name] = mutseq

print("Total time analyzing all the dataFrame {} s".format(time.time() - t0))

fail

temp_accnum = no_wild_type_repeated[['UniProt_ID','MUTATION','Tm_(C)','PDB_wild']]

arr_acc_temp = temp_accnum.to_numpy()

temp_dic={}

for i in arr_acc_temp:

    acc = i[0]

    if(i[1] == 'wild-type'):

        mut ='WT'

    else:

        mut = i[1].split('(')[0].strip()

    name="{}-{}".format(acc,mut)

    temp_dic[name]=[i[-2],i[-1].upper()]

with open('sequences_protherm.csv','w') as file:

    file.write("AccNumber,T_m,PDBID,Sequence \n")

    for k,v in Sequences.items():

        if(v==None):

            continue

        temp = temp_dic[k][0]

        code_pdb = temp_dic[k][1]

        text = "{},{},{},{} \n".format(k,temp,code_pdb,v)

        file.write(text)

import deepchem as dc

from rdkit import Chem

import pandas as pd 

Final_Data  = pd.read_csv("sequences_protherm.csv")

seq_list = list(Final_Data['Sequence '])

codes = ['A', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'K', 'L',

         'M', 'N', 'P', 'Q', 'R', 'S', 'T', 'V', 'W', 'Y']

max_seq= 0 

for i in seq_list:

    if(len(i)>max_seq):

        max_seq=len(i)

max_seq

OneHotFeaturizer = dc.feat.OneHotFeaturizer(codes,max_length=max_seq)

features = OneHotFeaturizer.featurize(seq_list)

temp = [float(x.split("(")[0]) for x in list(Final_Data['T_m'])]

dc_dataset = dc.data.NumpyDataset(X=features,y=temp)

dc_dataset.X.shape

from deepchem import splits

splitter = splits.RandomSplitter()

train, test  = splitter.train_test_split(dc_dataset,seed=42)

dc_dataset.X.shape[1:]

from tensorflow import keras

from tensorflow.keras import layers

model = keras.Sequential([

    layers.Dense(units=32, activation='relu', input_shape=dc_dataset.X.shape[1:]),

    layers.Dropout(0.2),

    layers.Dense(units=32, activation='relu'), 

    layers.Dropout(0.2),

    layers.Dense(units=32, activation='relu'), 

    layers.Dropout(0.2),

    layers.Dense(units=1),

])

model.compile(loss='mae', optimizer='adam')

print(model.summary())

history = model.fit(

    train.X, train.y,

    validation_data=(test.X,test.y),

    batch_size=24,

    epochs=10,

)

## perform a plot of loss vs epochs 

import matplotlib.pyplot as plt

history_df = pd.DataFrame(history.history)

history_df[['loss', 'val_loss']].plot()

dc_model = dc.models.KerasModel(model,dc.metrics.mae_score)

from deepchem import splits

splitter = splits.RandomSplitter()

train, test  = splitter.train_test_split(dc_dataset,seed=42)

train.X.shape

from sklearn.ensemble import RandomForestRegressor

from deepchem.utils.evaluate import Evaluator

import pandas as pd

seed = 42 # Set a random seed to get stable results

sklearn_model = RandomForestRegressor(n_estimators=100, max_features='sqrt')

sklearn_model.random_state = seed

model = dc.models.SklearnModel(sklearn_model)

model.fit(train)

from propy import PyPro

aaComplist = []

CTDList =[]

for seq in seq_list:

    Obj = PyPro.GetProDes(seq)

    aaComplist.append(np.array(list(Obj.GetAAComp().values())))

    CTDList.append(np.array(list(Obj.GetCTD().values())))

dc_dataset_aacomp = dc.data.NumpyDataset(X=aaComplist,y= temp)

dc_dataset_ctd = dc.data.NumpyDataset(X=CTDList,y= temp)

from deepchem import splits

splitter = splits.RandomSplitter()

train, test  = splitter.train_test_split(dc_dataset_aacomp,seed=42)

from sklearn.ensemble import RandomForestRegressor

from deepchem.utils.evaluate import Evaluator

import pandas as pd

print("RandomForestRegressor")

seed = 42 # Set a random seed to get stable results

sklearn_model = RandomForestRegressor(n_estimators=100, max_features='sqrt')

sklearn_model.random_state = seed

model = dc.models.SklearnModel(sklearn_model)

model.fit(train)

metric = dc.metrics.Metric(dc.metrics.mae_score)

train_score = model.evaluate(train, [metric])

test_score = model.evaluate(test, [metric])

print("Train score is : {}".format(train_score))

print("Test score is : {}".format(test_score))

print("SupportVectorMachineRegressor")

from sklearn.svm import SVR

svr_sklearn = SVR(kernel="poly",degree=4)

svr_sklearn.random_state = seed 

model = dc.models.SklearnModel(svr_sklearn)

model.fit(train)

metric = dc.metrics.Metric(dc.metrics.mae_score)

train_score = model.evaluate(train, [metric])

test_score = model.evaluate(test, [metric])

print("Train score is : {}".format(train_score))

print("Test score is : {}".format(test_score))

from deepchem import splits

splitter = splits.RandomSplitter()

train, test  = splitter.train_test_split(dc_dataset_ctd,seed=42)

from sklearn.ensemble import RandomForestRegressor

from deepchem.utils.evaluate import Evaluator

import pandas as pd

print("RandomForestRegressor")

seed = 42 # Set a random seed to get stable results

sklearn_model = RandomForestRegressor(n_estimators=100, max_features='sqrt')

sklearn_model.random_state = seed

model = dc.models.SklearnModel(sklearn_model)

model.fit(train)

metric = dc.metrics.Metric(dc.metrics.mae_score)

train_score = model.evaluate(train, [metric])

test_score = model.evaluate(test, [metric])

print("Train score is : {}".format(train_score))

print("Test score is : {}".format(test_score))

print("SupportVectorMachineRegressor")

from sklearn.svm import SVR

svr_sklearn = SVR(kernel="poly",degree=4)

svr_sklearn.random_state = seed 

model = dc.models.SklearnModel(svr_sklearn)

model.fit(train)

metric = dc.metrics.Metric(dc.metrics.mae_score)

train_score = model.evaluate(train, [metric])

test_score = model.evaluate(test, [metric])

print("Train score is : {}".format(train_score))

print("Test score is : {}".format(test_score))

%% Cell type:markdown id: tags:

# Tutorial 30: Protein Deep Learning

In this tutorial we will explore differents featurization of protein sequences, including one hot encoders, aminoacids composition and 3D contact maps. We will use some tools of DeepChem and additional packages to create a model to predict melting temperature of proteins ( a good measurement of protein stability )

In this tutorial we will  compare featurization of protein sequences such as one hot encoders and aminoacids composition. We will use some tools of DeepChem and additional packages to create a model to predict melting temperature of proteins ( a good measurement of protein stability )

the Melting temperature (MT) of a protein is a measurement of protein stability. This measure could vary from a big variety of experimental conditions, however, databases as  ProthermDB https://web.iitm.ac.in/bioinfo2/prothermdb/index.html contains a lot of thermodynamicall information of proteins and therefore a a big resource for the study of protein stability. Other information related with protein stability could be the change in Gibbs Free Energy $ \Delta \Delta G°$ due to a mutatation.

The study of protein stability is important in areas such as protein engineering and biocatalysis

%% Cell type:markdown id: tags:

# TODO

- include colab installer

- improve Melting temperature description and importance

- include 3D contact map

# Setup

%% Cell type:markdown id: tags:

To run DeepChem within Colab, you'll need to run the following installation commands. This will take about 5 minutes to run to completion and install your environment. You can of course run this tutorial locally if you prefer. In that case, don't run these cells since they will download and install Anaconda on your local machine.

%% Cell type:code id: tags:

``` python

!curl -Lo conda_installer.py https://raw.githubusercontent.com/deepchem/deepchem/master/scripts/colab_install.py

import conda_installer

conda_installer.install()

!/root/miniconda/bin/conda info -e

```

%% Cell type:code id: tags:

``` python

!pip install --pre deepchem

```

%% Cell type:code id: tags:

``` python

!pip install propy3

```

%% Cell type:markdown id: tags:

# Melting temperature  MT

the MT of a protein is a measurement of protein stability. This measure could vary from a big variety of experimental conditions, however, databases as  ProthermDB https://web.iitm.ac.in/bioinfo2/prothermdb/index.html contains a lot of thermodynamicall information of proteins and therefore a a big resource for the study of protein stability

# Dataset Extraction

%% Cell type:code id: tags:

``` python

#imports

import numpy as np

import pandas as pd

import matplotlib.pyplot as plt

```

%% Cell type:markdown id: tags:

The file all_measuremnetsProtherDB.tsv was extracted from the ProThermDB selecting a query of $T_m$ in the range of -52 and 220 °C. Additionaly the experiment conditions and Uniprot AccesionNumber were also requested

%% Cell type:code id: tags:

``` python

data =pd.read_csv("./all_measurementsProtherDB.tsv",delimiter="\t")

```

%% Cell type:code id: tags:

``` python

data.shape

```

%% Output

    (15629, 17)

%% Cell type:markdown id: tags:

# Data only with Measurement of Differential Scanning Calorymetry (DSC)

Data only with Measurement of Differential Scanning Calorymetry (DSC)

%% Cell type:code id: tags:

``` python

df = data[ (data['MEASURE'] == 'DSC') & (data['MUTATION'] == 'wild-type')]

```

%% Cell type:code id: tags:

``` python

df.shape

```

%% Output

    (3310, 17)

%% Cell type:code id: tags:

``` python

no_mutants_dataset = df.drop_duplicates(['PDB_wild'])

```

%% Cell type:code id: tags:

``` python

no_mutants_dataset.shape

```

%% Output

    (298, 17)

%% Cell type:markdown id: tags:

# Separate the only entries performed with Differential Scanning calorimetry

Separate the only entries performed with Differential Scanning calorimetry

%% Cell type:code id: tags:

``` python

df2 = data[ (data['MEASURE'] == 'DSC')]

```

%% Cell type:code id: tags:

``` python

df2.shape

```

%% Output

    (5797, 17)

%% Cell type:markdown id: tags:

# ProThermDB contains wild type entries of the same protein with differente value of melting temperature. In this tutorial we will keep the first wild type entry. ( Variations in $T_m$ could be caused by buffer selection or pH )

ProThermDB contains wild type entries of the same protein with differente value of melting temperature. In this tutorial we will keep the first wild type entry. ( Variations in $T_m$ could be caused by buffer selection or pH )

%% Cell type:code id: tags:

``` python

no_wild_type_repeated  =df2.drop_duplicates(['MUTATION','PDB_wild'],keep='first')

```

%% Cell type:code id: tags:

``` python

no_wild_type_repeated.sort_values('UniProt_ID')[['UniProt_ID','PDB_wild','PROTEIN']]

```

%% Output

          UniProt_ID PDB_wild            PROTEIN

    14049          -     3NA9  Fab15 light chain

    14116          -     3HC0     anti-LTbR scFv

    14115          -     3HC0     anti-LTbR scFv

    14114          -     3HC0     anti-LTbR scFv

    14113          -     3HC0     anti-LTbR scFv

    ...          ...      ...                ...

    7451      R9S082     1BVC          Myoglobin

    7452      R9S082     1BVC          Myoglobin

    7453      R9S082     1BVC          Myoglobin

    7439      R9S082     1BVC          Myoglobin

    7448      R9S082     1BVC          Myoglobin

    [1443 rows x 3 columns]

%% Cell type:code id: tags:

``` python

Mutation_list = no_wild_type_repeated[['UniProt_ID','MUTATION']].set_index('UniProt_ID')

```

%% Cell type:code id: tags:

``` python

Mutation_list

```

%% Output

                                        MUTATION

    UniProt_ID

    P0A877       P28S (Based on UniProt and PDB)

    P0A877       S33L (Based on UniProt and PDB)

    P0A877                             wild-type

    P0A877      M101V (Based on UniProt and PDB)

    P0A877      M101T (Based on UniProt and PDB)

    ...                                      ...

    P0AA25                             wild-type

    O00189      D190A (Based on UniProt and PDB)

    O00189      R283D (Based on UniProt and PDB)

    O49003                             wild-type

    O49003      C450A (Based on UniProt and PDB)

    [1443 rows x 1 columns]

%% Cell type:markdown id: tags:

# based on the UniprotAcc Numbers, we will use the python request module to access information of sequence in each entry

based on the UniprotAcc Numbers, we will use the python request module to access information of sequence in each entry

%% Cell type:code id: tags:

``` python

import requests, sys

def Seq_From_AccNum(num):

    '''

    perform a http  request to the EBI-API to obtain the sequence based on the Uniprot Accession number

    return a string

    '''

    requestURL = "https://www.ebi.ac.uk/proteins/api/features/{}".format(num)

    r = requests.get(requestURL, headers={ "Accept" : "application/json"})

    if not r.ok:

        print("Failure in Uniprot request")

        return None

        r.raise_for_status()

        sys.exit()

    responseBody = r.json()

    return responseBody['sequence']

```

%% Cell type:code id: tags:

``` python

import re

def MutateSeq(seq,Mutant):

    '''

    mutate a sequence based on a string (Mutant) that has the notation :

    A###B where A is the wildtype aminoacid ### the position and B the mutation

    '''

    aalist = re.findall('([A-Z])([0-9]+)([A-Z])', Mutant)

    #(len(aalist)==1):

    newseq=seq

    listseq=list(newseq)

    for aas in aalist:

        wildAA = aas[0]

        pos = int(aas[1]) -1

        if(pos >= len(listseq)):

            print("Mutation not in the range of the protein")

            return None

        MutAA = aas[-1]

        if(listseq[pos]==wildAA):

            listseq[pos]=MutAA

        else:

            print("WildType AA does not match")

            return None

    return("".join(listseq))

```

%% Cell type:code id: tags:

``` python

Mutation_list.sort_values('UniProt_ID',inplace=True)

```

%% Cell type:code id: tags:

``` python

Mutation_list

```

%% Output

                                                         MUTATION

    UniProt_ID

    -           3na9_H:V34I 3na9_H:G35S 3na9_H:E95Q 3na9_L:S95...

    -                                  3HC0_A:V56G (Based on PDB)

    -                                  3HC0_A:V68L (Based on PDB)

    -                                  3HC0_A:V68I (Based on PDB)

    -                                   3hc0_H:Q3G (Based on PDB)

    ...                                                       ...

    R9S082                                L70A (Based on UniProt)

    R9S082                               L136I (Based on UniProt)

    R9S082                               L136M (Based on UniProt)

    R9S082                               I112A (Based on UniProt)

    R9S082                                L70I (Based on UniProt)

    [1443 rows x 1 columns]

%% Cell type:markdown id: tags:

# The next cell will download from the EBI-API the sequence and perform all the reported mutations by ProthermDB that are consistent with Uniprot sequence .

The next cell will download from the EBI-API the sequence and perform all the reported mutations by ProthermDB that are consistent with Uniprot sequence .

%% Cell type:code id: tags:

``` python

import time

t0 = time.time()

Sequences = {}

fail  = {}

AccNum = []

count = 0

print("Perfoming data curation : ")

for accnumber, row in Mutation_list.iterrows():

    #if(count == 100):

    #    break

    #count += 1

    mutation = row['MUTATION'].split("(")[0].strip()

    print("{} - {}".format(accnumber,mutation), end =" ")

    name ="{}-{}".format(accnumber,mutation)

    if(accnumber=='-'):

        fail[accnumber] = [mutation]

        continue

    if accnumber not in AccNum:

        AccNum.append(accnumber)

        seq = Seq_From_AccNum(accnumber)

        if(seq == None):

            continue

    if(mutation =='wild-type'):

        name ="{}-{}".format(accnumber,"WT")

        Sequences[name]=seq

    else:

        mutseq = MutateSeq(seq,mutation)

        if(mutseq==None):

            if(accnumber not in fail ):

                fail[accnumber] = [mutation]

            else:

                fail[accnumber].append(mutation)

        Sequences[name] = mutseq

print("Total time analyzing all the dataFrame {} s".format(time.time() - t0))

```

%% Output

    Perfoming data curation :

    - - 3na9_H:V34I 3na9_H:G35S 3na9_H:E95Q 3na9_L:S95P - - 3HC0_A:V56G - - 3HC0_A:V68L - - 3HC0_A:V68I - - 3hc0_H:Q3G - - 3hc0_H:Q3D - - 3hc0_H:Q3V - - wild-type - - 3hc0_H:S16Q 3hc0_L:S46L - - 3hc0_H:Q3S - - 3na9_H:V34I 3na9_H:G35S - - 3na9_H:F50R - - 3na9_H:A60S 3na9_H:Q66H - - 3na9_H:V34I 3na9_H:G35S 3na9_H:E95Q - - wild-type - - 3hc0_H:M48I - - 3na9_H:V34I 3na9_H:G35S 3na9_H:F50R 3na9_H:A60S 3na9_H:Q66H - - 3na9_H:V34H - - 3hc0_H:M48G - - 3hc0_H:S16E 3hc0_L:S46L - - - - - wild-type - - 3hc0_L:S46L - - 3hc0_H:V20I - - 3hc0_H:S16E - - 3hc0_H:S16Q - - 3hc0_H:Q3A - - wild-type - - - - - 3na9_H:V34I 3na9_H:G35S 3na9_H:F50R 3na9_H:E95Q 3na9_H:A60S 3na9_H:Q66H A0A1D5PBP6 - wild-type A0QWG6 - wild-type A4QUT2 - Y273F A4QUT2 - M299A A4QUT2 - wild-type A4QUT2 - W140F B8YLY0 - S227G Mutation not in the range of the protein

    B8YLY0 - wild-type D0WVP7 - L272R D0WVP7 - L261Q D0WVP7 - L272P O00189 - D190A O00189 - R283D O15350 - wild-type O25103 - wild-type O25776 - wild-type O26652 - wild-type O26652 - D54R O26652 - R85E O26652 - A159E O26652 - Y102A O26652 - W138A O26652 - K143A O49003 - wild-type O49003 - C450A O58720 - wild-type O59170 - wild-type O66037 - wild-type O66037 - wild-type O66529 - wild-type O68541 - A26C A334V G348D L380C Mutation not in the range of the protein

    O68541 - A334V G348D O68541 - G348D O68541 - A334V O68541 - A26C L380C Mutation not in the range of the protein

    O74035 - D135A O74035 - D105A O74035 - E8Q O74035 - E8A O74035 - D7N O74035 - wild-type P00004 - wild-type P00044 - T102A C108T P00044 - C108A P00044 - N58I P00044 - G12A N58I P00044 - N58Q P00044 - N58S P00044 - N58H P00044 - N58T P00044 - N58A P00044 - N58V P00044 - N58M P00044 - N58L P00044 - F88Y L91A P00044 - wild-type P00044 - C108T P00044 - F88Y P00044 - T75E C108T P00044 - L91A P00044 - T75E T102A C108T P00099 - F56Y E65Y V100I P00099 - wild-type P00099 - F29A V35M F56Y E65Y V100I P00099 - F29A V35M P00099 - F56Y E65Y P00099 - V100I P00099 - F29A V35M F56Y E65Y P00099 - F29A V35M V100I P00169 - wild-type P00173 - R31H E36S WildType AA does not match

    P00183 - wild-type P00183 - T102V P00257 - H114Q P00257 - wild-type P00257 - H114R P00257 - T112S P00257 - T112A P00257 - Y140S P00257 - D134E P00257 - Y140L P00257 - Y140F P00257 - H114T P00257 - Y140W P00257 - C153S P00282 - wild-type P00282 - C23A C46A P00282 - D43E K148R P00282 - K44R P00282 - D43E K148R K44R P00282 - D82C K94C P00289 - wild-type P00362 - C152S P00362 - wild-type P00362 - N315T P00362 - H179N P00362 - Y285V P00362 - D284G P00362 - T36Q T40S L44Q P00362 - Y47G S49G P00362 - Y47G R53G P00362 - W312F P00383 - wild-type P00396 - wild-type P00439 - wild-type P00441 - C7A P00441 - C7A C112S P00441 - wild-type P00441 - C112S P00442 - wild-type P00560 - wild-type P00563 - wild-type P00590 - wild-type P00644 - E155G P00644 - D159G P00644 - G170V P00644 - E155G E157G P00644 - E157G D159G P00644 - E155G E157G D159G P00644 - V148L P00644 - E157V P00644 - D159A P00644 - V148A P00644 - wild-type P00644 - V148W P00644 - L219A P00644 - G170W P00644 - E157G P00644 - E157A P00644 - A151T P00644 - K106G P00644 - V105A P00644 - L89A P00644 - G161S P00644 - L107A P00644 - A172S P00644 - E155G D159G P00644 - H206L P00644 - V148L G170V P00644 - V148L G161S G170V P00648 - S132C H149C P00648 - D55A P00648 - D59G P00648 - wild-type P00648 - D69M P00648 - D101N P00648 - R130Q P00648 - I51A I98V P00648 - S139A P00648 - T73G P00648 - T63R P00648 - I98V P00648 - I143V P00648 - R116M P00648 - I135V P00648 - S138A P00648 - G100A P00648 - G99A P00648 - G100V P00648 - G99V P00649 - wild-type P00651 - D102S P00651 - F72E WildType AA does not match

    P00651 - F72K WildType AA does not match

    P00651 - H66T W85Y P00651 - H118A P00651 - W85Y H118A P00651 - A30C G114R V144C WildType AA does not match

    P00651 - G114R P00651 - A30C C91A C98A G114R V144C WildType AA does not match

    P00651 - C91A C98A G114R WildType AA does not match

    P00651 - C66A G114R C121A WildType AA does not match