Figure 4: Partial least squares analysis of the mechanistic model. | Scientific Reports

Figure 4: Partial least squares analysis of the mechanistic model.

From: Distinguishing autocrine and paracrine signals in hematopoietic stem cell culture using a biofunctional microcavity platform

Figure 4

(A) Model coefficients and corresponding VIP values were calculated for the data obtained from the 7 day cultures of HSPCs on top of TCP and PDMS-FN, PEG-HEP, or PEG-PEG microcavity arrays using 10 ng·mL−1 background cytokines in the medium. The dimensions of the model were reduced by deleting all variables with variable importance in projection (VIP) scores < 0.8. The simplified model was compared to the standard PLS fit model concerning the obtained cell numbers. The simplified model with fitted coefficients was validated by application to the set of data obtained from similar HSPC experiments, but at 30 ng background cytokine levels. Dashed lines of both graphs illustrate the 90% prediction range. (B) Plot shows the calculated model coefficients for centered and scaled data as a function of VIP. Coefficients with small values, close to 0, and VIP scores < 0.8 have no significant influence on the explanation of the y-response. The abbreviations ‘auto.’ and ‘para.’ represent autocrine and paracrine contributions in the model, respectively.

Back to article page