351
additional procedures are necessary for reliable determination of the mixture components. As can be deduced from table 12.6, the initial attempts presented above
are quite encouraging. For the four samples consisting of five-component mixtures,
in two cases all the components were correctly identified without false positives
and negatives. the remaining two cases gave only one false positive, but all the
components present were correctly determined. the identification was 100 % correct, with zero false positives and negatives for all five cases of three-component
mixtures. Finally, for an extremely complicated case of a mixture containing as
many as eight amino acids, seven could be identified. the only false negative was
isoleucine. Failure of its detection may be due to a simultaneous presence of leucine, which the results of the similarity analysis place very close to isoleucine. Still,
a more careful choice of the spectral interval could probably lead to the analytical
separation of these two amino acids, as can be inferred from Figs. 12.4 and 12.6.
A more general procedure can be envisaged. For example, first, we could define
different subregions of wavenumbers for each pair of template spectra, by optimizing a comparison measure. then, after an initial choice of components using the
algorithm described above we could focus on those with small scaling coefficients
in the reconstructed spectrum of the mixture. In the next step, we could check which
spectra of the remaining, non-identified amino acids are the most similar to the
chosen ones in the whole range of wavenumbers, where the analysis was performed
initially. Finally, we could substitute successively the spectra of the selected amino
Table 12.6 Identification results by the Canberra distance of second order derivatives of measured
mixture spectra denoted by letters; see table 12.4. All amino acids in the second column were
identified correctly using mechanical stop condition of the iterative algorithm (see Sect. 12.3.4)
mixture
Identified correctly
False positives
False negatives
a
Phe Ala
b
his Arg Pro
c
tyr Asn
d
gly thr Ser gln Leu
e
trp glu gln Ile
f
Ser thr Arg Ala
g
met Ala his gly Leu
h
glu Leu Ile
i
his gly Leu
j
glu met Lys
k
his gly tyr Pro
l
thr Ala
m
Asn
n
met val Leu Ile
glu
o
Phe gln
p
tyr thr Pro Asn Asp gly
q
gly glu Ala
r
tyr trp his Arg Ser
thr
s
trp Asp
thr Phe
t
tyr Cys Phe his Ser
Leu thr
Ile
12 Raman Spectra of Solid Aminoacids: Spectral Correlation Analysis …
additional procedures are necessary for reliable determination of the mixture components. As can be deduced from table 12.6, the initial attempts presented above
are quite encouraging. For the four samples consisting of five-component mixtures,
in two cases all the components were correctly identified without false positives
and negatives. the remaining two cases gave only one false positive, but all the
components present were correctly determined. the identification was 100 % correct, with zero false positives and negatives for all five cases of three-component
mixtures. Finally, for an extremely complicated case of a mixture containing as
many as eight amino acids, seven could be identified. the only false negative was
isoleucine. Failure of its detection may be due to a simultaneous presence of leucine, which the results of the similarity analysis place very close to isoleucine. Still,
a more careful choice of the spectral interval could probably lead to the analytical
separation of these two amino acids, as can be inferred from Figs. 12.4 and 12.6.
A more general procedure can be envisaged. For example, first, we could define
different subregions of wavenumbers for each pair of template spectra, by optimizing a comparison measure. then, after an initial choice of components using the
algorithm described above we could focus on those with small scaling coefficients
in the reconstructed spectrum of the mixture. In the next step, we could check which
spectra of the remaining, non-identified amino acids are the most similar to the
chosen ones in the whole range of wavenumbers, where the analysis was performed
initially. Finally, we could substitute successively the spectra of the selected amino
Table 12.6 Identification results by the Canberra distance of second order derivatives of measured
mixture spectra denoted by letters; see table 12.4. All amino acids in the second column were
identified correctly using mechanical stop condition of the iterative algorithm (see Sect. 12.3.4)
mixture
Identified correctly
False positives
False negatives
a
Phe Ala
b
his Arg Pro
c
tyr Asn
d
gly thr Ser gln Leu
e
trp glu gln Ile
f
Ser thr Arg Ala
g
met Ala his gly Leu
h
glu Leu Ile
i
his gly Leu
j
glu met Lys
k
his gly tyr Pro
l
thr Ala
m
Asn
n
met val Leu Ile
glu
o
Phe gln
p
tyr thr Pro Asn Asp gly
q
gly glu Ala
r
tyr trp his Arg Ser
thr
s
trp Asp
thr Phe
t
tyr Cys Phe his Ser
Leu thr
Ile
12 Raman Spectra of Solid Aminoacids: Spectral Correlation Analysis …