In my opinion, you should not use the peak analyzer option of origin 8.6, but rather calculate the second derivative directly (before smoothing your data) and then eventually smooth the second derivative if it is too noisy (but better is to measure with more scans to improve SNR):

Mark the two columns (wavenumber set as X, transmittance set as Y). Then: Analysis->Mathematics->Differentiate->open dialogue.

Visual inspection of the plotted result should allow you to determine the peak positions. If you cannot find any peak, your spectrum does not allow for amide I - analysis. In that case, there could be a problem with the sample.

You should be aware of the fact that you are looking for quite narrow peaks eventually, so it is important to measure the spectra with a resolution high enough to resolve these peaks (at least half of the expected peak width).

Just in case you need further reference on amide I analysis, I recommend lecture of the following papers:

• doi:10.1016/0006-291X(83)91016-1
• doi:10.1016/0968-0004(92)90305-S
• doi:10.1016/0003-2697(91)90155-M
• doi:10.1016/0079-6107(93)90006-6

To finally answer your questions:

Hope, this helps you!

Cheers, Jan

I'm happy I could help you with your analysis! It is really interesting to see the (unfortunately poor) quality of the 2nd derivative spectrum: As you will note, all values are multiples of 0.0025 absorption units. This indicates that you should improve the quality of your raw data: Could there be an issue with export from your spectrometer software or import in origin (e.g. are only the first two digits exported/imported or similar)? You need the full precision of your acquired data! Or could you measure for longer time to get a spectrum of better quality? In any case, it is essential that you calculate the derivative before smoothing the data. I think, some of your second derivative features could be true features, but some could also be artefacts due to the quantized nature of the (unsmoothed) spectrum

Since the peaks seem to be quite narrow, I don't expect the type of baseline correction to have a big impact on peak positions. Note however, that you could also have only three broad peaks around 1630, 1660, 1690 cm-1, respectively. For a quantitative analysis, it would be important to really understand the physical origin of the baseline taking substrate/host effects and particle scattering (by protein agglomerates) into account. This is certainly not an easy task to do and would require more information on substrate/host dielectric function, size distribution and so on...

Mr. Alavi,

To obtain an unambiguous information about secondary structure of protein by conventional IR-spectroscopy is a highly speculative task, generally.

Please pay attention to our series of papers, including a review article devoted to application of linear polarized IR-spectroscopy for structural analysis of small peptides and their transition metal complexes [Refs. 1-8,10]. There are shown here a small set of references, but they have contained the largest representative part of data obtained. Please take into consideration that this is, in fact, (polarized IR-spectroscopy) the unambiguous tool for experimental assignment of IR-bands in condense phase. We have correlated it also with X-ray diffraction data of a single crystal of a peptide in [Ref. 3].

But, despite it powerful ability, reliable information we have achieved studying oligopeptides, furthermore when corresponding amino acid residues are not so complicated from an IR-spectroscopic point of view, meaning molecular structural fragments yielding to complex IR-profile such as for example, Trp, Phe, Tyr, Arg, and related (Ref. 2 and 4).

Please pay attention to our paper also devoted to relatively short chain peptides such as bradykinin (Brad) and angiotesine (AngII) (Ref. 10, https://www.researchgate.net/publication/256731675_Coordination_ability_of_bradykinin_with_ZnII-_and_AgI-metal_ions__Experimental_and_theoretical_study), where are shown IR-spectra within amide I region. In addition you may find as attechment the corresponding polarized IR spectra (shown as s,p). In fact even applying mathematical data processing approaches such as deconvolution, curve fitting and more, there is unable unambiguously to assign the corresponding frequencies to expected set of molecular motions, even using labelling, there was not obtained a remarkable increasing in the analytical information about the structure of those derivatives. The amide I and II IR-bands are strong overlapped with increasing the number of the amino-acid sequences and the applicability of the chemometric approaches is strongly reduced. As you can see in the attechment there is difficult to assign even 10-12 amio acid residues. Please pay attention to Ref. 9, wher we have exactly validated the reducing procedure on concentional FTIR. There are shown examples on amino acids and small peptides.

Given that, a single crystal protein X-ray diffraction is the method of choice allowing to achieve structural information for proteins.

[Ref. 1] Ivanova, B. (2005), Solid-state linear-dichroic infrared spectroscopic analysis of the dipeptide S-Phe-S-Phe and its mononuclear Au(III) complex, Journal of Coordination Chemistry 58, 587-593

[Ref. 2] Ivanova, B. (2006) IR-LD spectroscopic characterization of l-Tryptophan containing dipeptides, Spectrochimica Acta - Part A 64, 931-938

[Ref. 3] Ivanova, B.; Arnaudov, M.; Todorov, S.; Sheldrick, W.S.; Mayer-Figge, H. (2006) Structural analysis of the tripeptide glycyle-methionyl-glycine (H-Gly-Met-Gly-OH) and its hydrochloride, Structural Chemistry 17, 49-56.

[Ref. 4] Ivanova, B. (2006) Stereo-structural and IR-spectral characterization of histidine containing dipeptides by means of solid-state IR-LD spectroscopy and ab initio calculations Journal of Molecular Structure 782, 122-129.

[Ref. 5] Kolev, T.; Koleva, B.; Zareva, S.; Spiteller, M. (2006) Au(III)-complexes of the alanyl-containing peptides glycylalanine and glycylalanylalanine - Synthesis, spectroscopic and structural characterization

Inorganica Chimica Acta 359, 4367-4376.

[Ref. 6] Koleva, B.; Kolev, T.; Spiteller, M. (2007) Spectroscopic and structural elucidation of l-tyrosine-containing dipeptides valyl-tyrosine and tyrosyl-alanine: Solid-state IR-LD spectroscopy, quantum chemical calculations and vibrational analysis Spectrochimica Acta - Part A 68, 1187-1196

[Ref. 7] Koleva, B.; Kolev, T.; Lamshoeft, M.; Spiteller, M. (2008) Synthesis, spectroscopic analysis and structure deduction of gold(III), palladium(II) and platinum(II) complexes with the tripeptide glycyl-l-phenylalanyl-glycine Transition Metal Chemistry 33, 911-919.

[Ref. 8] Kolev, T.; Spiteller, M.; Koleva, B. (2010) Spectroscopic and structural elucidation of amino acid derivatives and small peptides: experimental and theoretical tools, Amino Acids , 38, 45-50.

[Ref. 9] Ivanova, B.; Tsalev, D.; Arnaudov, M. (2006) Validation of reducing-difference procedure for the interpretation of non-polarized infrared spectra of n-component solid mixtures Talanta 69, 822-828.

[Ref. 10] Ivanova, B., Spiteller, M (2012) Coordination ability of bradykinin with ZnII- and AgI-metal ions - experimental and theoretical study Inorganica Chimica Acta, 392, 211-220.

Article Coordination ability of bradykinin with ZnII- and AgI-metal ...

Hello, 

I read all comments , I have same problem I could not interpret my results of FTIR. Could anyone give me all steps to have secondary structure of my protein in ORIGINPRO. I could not calculate secondary derivative. I have this artcile that shows all steps but I don't know how to do its In Origin. Please I need your Help

Let us consider a less complicated case of data-processing of IR-pattern of octapeptide (Attachment), before to comment the reliability of the data in ref. 1 (382 | VOL.10 NO.3 | 2015 | Nature protocols), even describing their qualitative character. Because of from a quantitative analytical point of view the data are completely unreliable.

So, page 1 (attachment) shows an IR-spectrum of an octapeptide. Within region 1680+/- 60 cm-1 there are expected 8 amide I modes (There are other bands as well as, but let us focus attention to Amide I bands)

Let us now reproduce step-by-step the data-processing shown in reference 1. Page 2 (attachment, here) contains the first and second derivatives of the original pattern, while page 3 reflects the smoothing and baseline corrections, which are carried out in order to obtain the corresponding representation of the spectra from Fig. 5 (Ref. 1).

Now, let us perform further a chemometric analysis, comparing quantitatively the output-data from curve-fitting of the unprocessed original spectrum (page 1) and from the corresponding second derivative analysis (Compare data in pages 4 and 5). There is obtained difference in 41 cm-1, which is a great deviation accounting for he fact that the resolution of the measurements is at about 0.5 cm-1. In addition, the analysis do not elicit the data by setting up the physical meaning.

However, the authors have written in Table 1 (Ref. 1) the following differences between a-helix and b-turn (12 cm-1) and between a-helix and b-sheet (18 cm-1). It is clear that both values are within the range of uncertainty of the protocol.

The next problem is that even-order differentiation, furthermore accompanied with filtering yields to a spectrum-like profile of the patterns. Always there is obtained a reduction of the band height and increasing its width. This causes for a decreasing in the resolution. In addition, there are generated side-lobes (Attachment, page 2). In particular when the profile consists of large number of overlapped bands, the procedure yields to increasing in the errors determining the spectral parameters. There is obtained a lost of relative band area relationships, as well. The latter statement is further illustrated looking at the 'A' values (Pages 4 and 5, attachment; A - denotes area). But the unreliable determination of the areas prevents a quantification of the spectra and determining of the absorption coefficients. In spite of the authors (Box 1, page 388, ref. 1) have claimed that the protocol provides even quantitative information as well (NB! There is a lack of chemomentric analysis in ref. 1. In this context claims/conclusions about quantitative or qualitative analyses are unreliable).

NB2! The data-processing steps in ref. 1 can be performed by Origin.

Is it necessary to do curve fitting with Origin?

why cant it be done with IR software.

can anyone explain?

Dear, 

I was dealing with the same problem, and I prepared an automatic theme for Origin that can be used to solve the secondary structure. Actually I'm using 11 peaks present in literature to study silk fibroin conformation. I found a lot of confusion on this field, and a lot of articles that treat peak fitting as deconvolution, you should be aware of the fact that peak fitting is not a deconvolution. So you can act in different ways:

1) Find the peaks with a secondary derivative and than fit them.

2) Deconvolve the signal to obtain the position of the peaks and then fit them (self-fourier deconvolution).

3) Assign the value of the peaks in Origin, setting constrain in the fit parameters window (I used to set the bandwidth in witch I expect to find each peaks), set also the a constrain on the area of each peaks  to be larger or equal to zero and then leave Origin to fit the data. The constrain on the area is important so Origin can put some peaks to 0 it the peak is not necessary to fit the data. 

I'm actually using the third method using Voigt peaks instead of Gaussian, but I'm thinking on working on the data directly using the second. There's no correct method to solve this problem and it is always a fit so be aware of this. If you fit the data in different ways you should have different results.