DIFFERENCES IN PROTEIN PROF ILES To investigate differences in protein profi les between samples, a multivariate data analysis approach was performed using a supervised partial least squares-discriminant analysis (PLS- DA). When comparing the protein profi les of the 217 quantifi ed proteins in each pool, the pooled samples belonging to the very curly hair group or the very straight hair group were able to be distinguished based on the protein abundance profi les of the proteins (Figure 1). To better understand which proteins were the key contributors in the discrimination between these two groups in the PLS-DA, the top 20 proteins that contributed the most to this discrimination in the fi rst component were determined. An overview of the protein groups they belong to is listed in Table I. Keratin 85 was the strongest contributor to the distinction of the very straight hair shaft pools, whereas KAP 13-2 had the most dis- crimination power in the very curly hair pooled samples. DIFFERENCES IN PROTEIN ABUN DANCE Next, a univariate analysis was performed to detect proteins whose abundance differed according to the level of curliness. Two different parameters in this statistical analysis are of major importance: (i) the p value: how signifi cantly different was the protein abundance Figure 1. Multivariate analysis via PLS-DA enables clustering of the 20 pooled samples measured in duplicate (A1 to T2) based on the protein abundance levels of 217 proteins present in the samples. A clear discrimination between the very curly hair group and the very straight hair group is seen. JOURNAL OF COSMETIC SCIENCE 254

difference between the two experimental groups, and (ii) the fold change: how large was the difference in protein abundance between the two groups. An overview of these data is rep- resented in the volcano plot in Figure 2. In this volcano plot, every dot represents a specifi c protein species. All gray dots are proteins that were not statistically signifi cantly different between the two experimental groups. The purple and yellow dots represent the proteins that passed the signifi cance threshold and are therefore of high interest. Each of these pro- teins had a signifi cant p value (so the proteins were signifi cantly different between the two experimental groups), and the proteins were also at least twofold higher or lower in abun- dance in a specifi c experimental group. The yellow dots represent proteins with higher abundances in the very straight samples, whereas the purple dots represent proteins that were higher in abundance in the very curly hair shape samples. Interestingly, 14 unique pr oteins in our dataset displayed a signifi cant difference in the abundance level (adjusted p-value d 0.05) between the very curly versus the very straight hair group with at least a twofold difference. Table II lists all the s ignifi cant proteins (most signifi cant protein listed fi rst), with their respective signifi cance level (adjusted p value via Benjamini–Hochberg correction) and fold change. Some proteins (e.g., K1) detected in these ethnically diverse samples are known to exist in several protein isoforms, and as these different isoforms are detected with the same peptides, only one of them is displayed in the table. Of these 14 proteins of interest, eight proteins have a higher abundance in the very straight hair samples, whereas six proteins have a higher abundance level in the very curly hair samples. Because the fold-change calculations were based on mean values across the different pools within the same experimental group, additional analysis of the variability of the protein abundances within and between these experimental groups can generate additional insights. In Figure 3, the abundance levels of the 14 proteins found to have signifi cantly different expression levels between the curly hair groups and the straight hair groups are represented. For each protein, the relative abundance levels for every pooled sample (thus 20 datapoints for very straight and 20 datapoints for very curly hair) are represented in a boxplot. The upper, middle, and lower lines represent the fi rst quantiles, medians, and third quantiles Table I Overview of the proteins most important in discriminating the two experimental groups based on PLS-DA Protein group Group contributor Importance K85 Very straight −0.1429 Methanethiol oxidase Very straight −0.1308 K34 Very straight −0.1299 K81 Very straight −0.1127 Protein S100-A3 Very straight −0.1122 Histone H2AX Very straight −0.1098 KAP 9-8 Very straight −0.1083 KAP 2-1 Very straight −0.1082 KAP 4-2 Very straight −0.1062 K38 Very curly 0.1069 K10 Very curly 0.1083 KAP 3-3 Very curly 0.1099 Protein band 4.1 Very curly 0.1169 KAP 4-4 Very curly 0.1170 KAP 13-2 Very curly 0.1192 HAIR SHAPE PROTEOMICS 255