The purpose of this comment is to reply to the one of J. Bak and T. M. Jørgensen [1] regarding our recent report on the analysis of human chromosomes using Raman spectroscopy [2]. The authors misunderstood the main purpose of our paper. As stated in the Abstract, the purpose of our presented work was to develop an approach (LTRS spectroscopy & manipulation plus G-banding) to obtain Raman spectral patterns for human chromosomes numbers 1, 2, and 3, which were verified by the staining-based G-banding technique after the procedures of trapping a chromosome, obtaining the spectrum, and manipulating/fixing the chromosome in another chamber of the slide. The use of new technology (LTRS & G-banding) allows us to obtain the spectra of human chromosomes numbers 1, 2, and 3, for the first time. To our point of view, this is significant technical progress because this report details the first time that the spectra of human chromosomes were obtained before staining (for verification purposes). After obtaining the spectra of these human chromosomes, we determined if these spectra could be used to distinguish human chromosomes 1, 2, and 3 from each other. We did not claim that these three chromosomes have been identified or differentiated completely by their spectra, although the preliminary results partially indicated this potential (in Fig. 4). In fact, in our conclusions, we stated that “it is possible that this LTRS system may be able to detect the minute differences between one person’s chromosome and another’s”. We should not expect that the problem of identifying human chromosomes can be completely solved in this pivotal work, which will require significantly more effort.

The problem of distinguishing human chromosomes by their Raman spectra is a difficult one, depending on many unknown factors, such as the variations among individual human donor chromosomes (due to differing genetic/allelic content), the inherent differences in biochemical compositions and structures of human chromosomes numbers 1, 2, and 3, the variations in the controlled conditions in culturing white blood cells and preparing human chromosomes, as well as the sensitivity and the reproducibility of the detection system. The contribution of our paper was to provide these spectral data (Fig. 2, Fig. 3 and Table 1) under the current experimental conditions and present a differential trial.

Bak and Jørgensens’ interpretation and understanding of the presented spectra data in Fig. 3(a), Fig. 3(b) and Table 1, upon which their critical comment is based, is incorrect. They thought that the statistical analysis presented in Fig. 3(b) and Table 1 show opposite conclusions, when compared to the difference spectra in Fig. 3(a). Actually, the data presented in Fig. 3(b) and Table 1 show both the mean values and standard deviation (the squared root of variance) of the peak height or the ratio between the heights of a few bands. However, the difference spectra in Fig. 3(a) shows the difference in the spectral shapes or profiles in wavenumber shifts. Therefore, Fig. 3(a) has a different meaning or shows a different spectral property from that of Fig. 3(b) and Table 1. Even though the three chromosome groups have similar values in peak heights or ratios in some specific bands on average, they may still have differences in small wavenumber shifts and spectral shapes. Indeed, the positive-negative patterns in the difference spectra in Fig. 3(a) shows the differences in small wavenumber shifts and spectral shapes. The purpose of Fig. 3(b) and Table 1 is to show the differences in peak height or ratio in certain bands in the three chromosome groups, and how much variation, if any, exists; whereas the purpose of Fig. 3(a) is to show that the three chromosome groups have differences in overall spectral profiles, which include small wavenumber shifts in the above peaks or bands. Their comment on Fig. 4 is also incorrect, for the same reasons described above.

Bak and Jørgensen state that the dispersive-like spectra in difference spectra in Fig. 3(a) was caused by the uncertainty in wavenumber axis. As stated in our original paper, the wavelength, Raman shift, and spectral response of the LTRS system was calibrated with polystyrene beads (reference [7] of our original paper). The wavenumber repeatability of the LTRS was tested with 1001 cm^-1 band of polystyrene beads to be below 1.0 cm^-1 although the minimum spectral bandwidth was ~6 cm^-1. The wavenumber shift that was a response to the dispersive-like difference chromosome spectra in Fig. 3(a) is relatively large compared to the wavenumber repeatability. These shifts cannot be caused by day-to-day wavenumber shifts of the LTRS system. As a direct test, we did not observe the dispersive-like spectral profile in the difference spectra at 1001 cm^-1 band between two identical non-biological polystyrene particles, even though they were measured on different days.

We also disagree with Bak and Jørgensen interpretation of the grouping of chromosome classes #1, #2, and #3 as the result of as incomplete experimental design. They regarded the sources of variances that contribute to the measurement as being due to day-to-day variations such as different operators of the equipment, daily calibration of the set-up, and grouping the samples representing different chromosomes before the measurements have taken place. This is unlikely because the data on the chromosomes in Fig. 4 were collected on 12 different days for Fig. 4(a) and on 6 different days for Fig. 4(b), while the data collected each day contain chromosome number 1 (group 1), number 2, and number 3. However, the chromosome numbers 1, 2, and 3 that were collected on different days were grouped in the appropriate clusters as described in the original paper. This cannot be explained as day-to-day variations.

Finally, we disagree with Bak and Jørgensens’ comment that there were missing details (such as what was the spectral range and how many data points were used for the analysis) for GDA analysis in our paper. In fact, in the text of our paper, we mentioned the generalized discriminate analysis (GDA) and have given one of our own references (reference [11] in our original paper), in which all the details were given, including the validation scheme that Bak and Jørgensen mentioned. In that reference [11] (C. Xie et al, “Identification of single bacterial cells in aqueous solution using confocal laser tweezers Raman spectroscopy,” Anal. Chem. 77, 4390-4397 (2005)), the analysis details used in the Opt. Exp paper were given: the spectral range of 600-1800 cm^-1 was used for multivariate analysis, which corresponds to 751 data points/spectrum.

In conclusion, while we agree that there were some inaccuracies in wording such as that the bars shown in Fig. 3(b) should be standard deviation (SD) and not variance (that is the square of SD), we disagree with their comments because Bak and Jørgensen misunderstood the original points of the paper, and also partially misunderstood the meanings of presented spectra data in Fig. 3(a), Fig. 3(b) and Table 1.