03_chmi_aim3.md

In mvazquezs/chmitools: Package to manage & obtain results of the CHMI Study

info for aim_3

data subset

• 1st filter: datasset == 'L1' (only LACHMI Study).

general summary

> dat_3 <- chmi.phen(
    data_type = 'ab_data',
    aim_data = 'aim_1',
    fold_change = FALSE,
    group_tr = 'ab_select')

> summary(dat_3) 

  original_id   dataset    gender     gr_hbs      gr2_hbs      hb_status_ori  ppp_time             status_ori         antigen       log10_mfi         mfi_corr        
 L1_002 : 610   L1:13542   F:6100   naive:2196   naive: 2196   AA:8418       Min.   :11.92   naive      : 2196   ama1_3d71: 666   Min.   :0.7193   Min.   :        5  
 L1_007 : 610              M:7442   AA   :6222   A_   :11346   AS:5124       1st Qu.:13.51   semi_immune:11346   ama1_fvo : 666   1st Qu.:3.5341   1st Qu.:     3421  
 L1_009 : 610                       AS   :5124                               Median :15.97                       cel_tos  : 666   Median :4.8415   Median :    69416  
 L1_010 : 610                                                                Mean   :16.88                       csp      : 666   Mean   :4.7306   Mean   :  9942372  
 L1_011 : 610                                                                3rd Qu.:19.00                       cy_rpa1  : 666   3rd Qu.:5.9715   3rd Qu.:   936567  
 L1_013 : 610                                                                Max.   :24.89                       cy_rpa2  : 666   Max.   :8.8091   Max.   :644245615  
 (Other):9882                                                                NA's   :4880                        (Other)  :9546                                       
  t_igg      t_point       t2_point   
 IgG :2331   C_1:2928   C_1    :2928  
 IgG1:2331   D13:2806   D11_D13:2806  
 IgG2:2331   D19:2318   D19    :2318  
 IgG3:2220   D28:2562   D28    :2562  
 IgG4:1998   D7 :2928   D7     :2928  
 IgM :2331     

demo to understand linear mixed models aim3

The demo contains the explanation to understand the procedure done in linear models for the aim 3. All examples have been done with the IgG istoype & AMA1_3D71 antigen. demo to understand the coefficient in linear mixed model (LMM). demo to understand the Simultaneous Tests for General Linear Hypotheses (GLHT).

summary() for the gr_hbs with 3 levels in LMM.

Linear mixed model fit by REML. t-tests use Satterthwaite's method ['lmerModLmerTest']
Formula: log10_mfi ~ t2_point_lm * gr_hbs + (t2_point_lm | original_id)
   Data: .x

REML criterion at convergence: 152.6

Scaled residuals:
    Min      1Q  Median      3Q     Max
-2.3791 -0.3007  0.0408  0.3286  3.1782

Random effects:
 Groups      Name        Variance  Std.Dev. Corr
 original_id (Intercept) 1.0655633 1.03226       
             t2_point_lm 0.0006115 0.02473  -0.73
 Residual                0.0601413 0.24524       
Number of obs: 111, groups:  original_id, 25

Fixed effects:
                     Estimate Std. Error       df t value Pr(>|t|)    
(Intercept)           4.48319    0.46975 21.95865   9.544 2.86e-09 ***
t2_point_lm           0.04272    0.01293 18.47458   3.305  0.00383 **
gr_hbsAA              2.99953    0.56645 21.94396   5.295 2.61e-05 ***
gr_hbsAS              3.63161    0.58712 22.11956   6.186 3.09e-06 ***
t2_point_lm:gr_hbsAA -0.01839    0.01542 18.20749  -1.192  0.24854    
t2_point_lm:gr_hbsAS -0.03487    0.01595 18.27883  -2.186  0.04207 *  
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Correlation of Fixed Effects:
            (Intr) t2_pn_ gr_hAA gr_hAS t2__:_AA
t2_point_lm -0.677                              
gr_hbsAA    -0.829  0.562                       
gr_hbsAS    -0.800  0.542  0.664                
t2_pnt_:_AA  0.568 -0.838 -0.685 -0.454         
t2_pnt_:_AS  0.549 -0.810 -0.455 -0.690  0.679

• Interpretation of fixed effects
• Intercept shows base mean of log10_mfi for the base level (naive) of the gr_hbs group, for the average of t2_point_lm.
• t2_point_lm shows increase or decrease in log10_mfi to naive group for every 1-unit increase in t2_point_lm.
• gr_hbsAA or gr_hbsAS shows increase or decrease in log10_mfi in comparison to naive group, for the average of t2_point_lm.

• t2_point_lm:gr_hbsAA or t2_point_lm:gr_hbsAS shows increase or decrease in log10_mfi (in AA or AS groups) in comparison to naive group for every 1-unit increase in t2_point_lm. To have this interpretation, we have to add up the interaction term plus t2_point_lm.

• Interpretation of fixed effects (example)

• Intercept tells us that the naive group will have 4.48 log10_mfi, for an average t2_point_lm.
• t2_point_lm naive group will increase in 0.04 lo10_mfi, for every 1-unit increase in t2_point_lm.
• gr_hbsAA or gr_hbsAS would expect to have 2.99 or 3.63 log10_mfi more than the naive group, for an average t2_point_lm.

• t2_point_lm:gr_hbsAA or t2_point_lm:gr_hbsAS AA or AS groups will increase in (0.04 + (-0.01)) or (0.04 + (-0.03)) log10_mfi, in comparison to naive group for every 1-unit increase in t2_point_lm.

• General interpretation

• The increase in log10_mfi for having a higher t2_point_lm differs among groups.

summary() for the gr_hbs with 3 levels in glht()

 Simultaneous Tests for General Linear Hypotheses

Fit: lmer(formula = as.formula(.y), data = .x)

Linear Hypotheses:
              Estimate Std. Error z value Pr(>|z|)    
t2_point_lm == 0  0.04272    0.01293   3.305  0.00095 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
(Adjusted p values reported -- single-step method)

• General interpretation
• The slope t2_point_lm (corresponding to naive group) is significantly different from zero.
• The test interpretation for the t2_point_lm:gr_hbsAA & t2_point_lm:gr_hbsAA interactions is likewise.

Bibliography

See nexts tutorials to obtain more information about the interpretation of the interaction terms in categorical and continuous data. tutorial_01 tutotial_02

mvazquezs/chmitools documentation built on May 1, 2020, 2:06 a.m.