Algorithm 1 The algorithm for transformation of SMILES strings into SMILES matrices
Require: $\mathbb{D}$	▹ SMILES dataset
$\mathbb{U}\leftarrow Ø$	▹ Unique symbols
$m\leftarrow 0$	▹ Maximal symbol length
$s\leftarrow ""$	▹ Empty symbol
for each $\mathrm{SMILES}\in \mathbb{D}$ do
for each $s\in \mathrm{SMILES}$ do
if $s\notin \mathbb{U}$ then
${\mathbb{U}}^{⌢}s$
end if
if $dim\left(\mathrm{SMILES}\right)>m$ then
$m\leftarrow dim\left(\mathrm{SMILES}\right)$
end if
end for
end for
$T_{dim\left(\mathbb{U}\right)\times m}$	▹ One-hot matrix to be filled
$\mathbb{T}\leftarrow Ø$	▹ Transformed dataset
$P\leftarrow Ø$	▹ MP values from dataset
$y\leftarrow 0$	▹ Output value
for each $\mathrm{SMILES}\in \mathbb{D}$ do
$i\leftarrow 0$	▹ For tracking vector columns
for each $s\in \mathrm{SMILES}$ do
$P\leftarrow {\mathbb{D}}_P$
$y\leftarrow {\mathbb{D}}_y$
if $s\in \mathbb{U}$ then
$\mathrm{Let}x\mathrm{be}\mathrm{such}\mathrm{that}{\mathbb{U}}_x=s$
$T_{x,i}=1$
$\forall x^{\prime }\in dim\left(\mathbb{U}\right)\wedge x^{\prime }\ne x\Rightarrow T_{x^{\prime },i}=0$
else
for each $xin[1,dim(\mathbb{U}\left)\right]$ do
$T_{x,i}=0$
end for
end if
$i\leftarrow i+1$
$\mathbb{T}\leftarrow (P,T,y)$
end for
end for