Processed Raman spectra of bone mineralization in osteopetrosis murine models (oc/oc and Rankl-/-) and wild-type controls

Dataset description This dataset accompanies the : Ventura et al. “Subtype-Specific Bone Mineralization Defects and Early Treatment Amelioration in Murine Models of Autosomal Recessive Osteopetrosis Revealed by Raman Spectroscopy.” Bone (2026). https://doi.org/10.1016/j.bone.2026.117989 The dataset contains Raman spectra acquired from intact, non-decalcified ex vivo murine bones. The study was designed to investigate bone compositional alterations in autosomal recessive osteopetrosis (ARO), comparing healthy wild-type (WT) mice with two osteopetrotic models: the severe TCIRG1-deficient oc/oc model and the milder RANKL-deficient Rankl-/- model. The dataset also includes Raman spectra from WT mice at different postnatal ages, collected to evaluate physiological changes in bone composition during early postnatal growth, and spectra from oc/oc mice treated with hematopoietic stem cell transplantation (HSCT/BMT), analyzed at postnatal day 18 to assess early treatment-associated changes. Raman spectra were collected from skull and long bones. Skull measurements were performed on the cortical surface of parietal bones, whereas long bone measurements were performed on the proximal metaphysis of femurs and/or tibiae. When both femur and tibia were available, spectra were pooled under the “long bone” category. For each mouse and bone, 40 single Raman spectra were collected. The dataset can be used to reproduce or extend the analyses reported in the manuscript, including:(i) comparison of Raman-derived compositional parameters among WT, oc/oc, and Rankl-/- mice at postnatal day 14;(ii) analysis of genotype- and severity-associated spectral differences between mild and severe osteopetrosis models;(iii) evaluation of age-related changes in WT bone composition during postnatal growth; and(iv) assessment of early mineral-to-matrix changes in transplanted oc/oc mice compared with untreated oc/oc and WT controls. The spectra can be processed to extract standard Raman-derived bone compositional parameters, including mineral-to-matrix ratio, mineral crystallinity, and carbonate substitution. These parameters are based on Raman bands associated with the mineral and organic matrix components of bone, such as phosphate, carbonate, and amide/collagen-related bands. Raman measurements were performed using a home built Raman microscope with 660 nm laser excitation. Bones were fixed, stored in ethanol, rehydrated/immersed in distilled water during acquisition, and measured without decalcification or sectioning. Spectra were acquired from randomly selected points on the cortical bone surface using constant acquisition settings across samples. Data collection The complete dataset includes Raman spectra from a total of 47 mice, for an overall number of 3200 spectra. Raman measurements were performed using a home-built confocal Raman microscope with 660 nm laser excitation. Bones were fixed, stored in ethanol, rehydrated/immersed in distilled water during acquisition, and measured without decalcification or sectioning. Spectra were acquired from randomly selected points on the cortical bone surface using constant acquisition settings across samples: a 60×/1.2 NA water-immersion objective, 25 mW excitation power at the sample surface, 1 s exposure time, and five averaged acquisitions per spectrum. For each mouse and bone, 40 spectra were collected, with the laser focused approximately 3–5 µm below the bone surface. Calibration and preprocessingWavenumber calibration was performed using toluene and an ArHg lamp as references, while intensity calibration was performed using a calibrated white lamp. Spectra were preprocessed using a custom Python script based on NumPy, Pandas, and SciPy. The preprocessing workflow included cosmic ray removal using a Modified Z-score approach, baseline correction by asymmetric least squares (AsLS) to reduce sample autofluorescence and quartz background, interpolation on a regular Raman shift grid from 380 to 3050 cm⁻¹ with 1 cm⁻¹ spacing using cubic spline interpolation, and unit-vector normalization to reduce intensity differences among spectra. Spectra included in this dataset are provided after calibration and preprocessing. Raman-derived parametersMineral-to-matrix ratio, carbonate-to-phosphate ratio, and crystallinity can be calculated from the processed spectra using the phosphate band around 960 cm⁻¹, the carbonate band around 1070 cm⁻¹, and the amide I band around 1660 cm⁻¹, as described in the associated manuscript. Data structure Each column corresponds to a single Raman spectrum of the dataset. The first rows contain metadata related to the measured sample, as indicated in the first column: Spectrum: identifier for the spectrum, indicated as RV[animal number]_[n]. n ranges from 1 to 40 since, for every mouse and bone, 40 single spectra were collected. Units: unit of measurement of the spectrum intensity. “A.u.” stands for arbitrary units. Bone: bone from which the spectrum was collected. Possible values are skull, femur, and tibia. Bone type: bone category used for the analyses. Possible values are skull and long bone, the latter including both tibia and femur. Pathology: genotype or experimental group of the mouse. Possible values are WT, oc/oc, Rankl-/-, and HSCT. Sample: animal identifier, indicated as RV[animal number]. Age: age of the animal, expressed in days. Possible values are 7, 14, 18, 21, 28, and 35. Sex: sex of the animal, male (m) or female (f). All following rows contain Raman intensity values for each spectrum. The corresponding Raman shift is reported in the first column and is expressed in cm⁻¹.