Three groups rejection, postrejection and nonrejection comparison for renal transplant patients

1
Three groups (rejection, post-rejection and
non-rejection) comparison for renal transplant
patients

• Xuelian Wei
• Statistics, UCLA

2
Sampling

• For each patient, after transplantation, the
  blood plasma samples will be collected
  periodically.
• If one patient has been suspected to have a
  rejection, a biopsy diagnosis was applied.
• If the result is negative(non-rejection), the
  sample near the biopsy date is treated as
  non-rejection sample (group C).
• If the result if positive(rejection), the sample
  near the biopsy date is treated as rejection
  sample (A). Immunosuppression treatment applied,
  the sample collected around one month later is
  treated as post-rejection sample (group B).
• Sample collected under different condition (even
  from the same patient) has a unique sample_id.
• 7 patients have all 3 groups data.

3
Sampling...

• 18 patients in AB gt 25 pseudo patients.
• 12 patients had only one sample pair collected
  and profiled.
• 2 patients has only one sample pair collected but
  profiled in two chips. (same sample_id, technical
  repeats).
• 3 patients had two sample pairs collected and
  profiled. (different sample_id, biological
  repeats).
• 1 patients had two sample pairs collected and
  profiled three times.

4
Sampling

• 49 patients in C gt 56 sample_ids.
• 7 patients are also show up in group AB.
• 7 patients have two sample_ids each.
• Two different assumptions
• I. Sample collected on different time or profiled
  on different chip (even from the same patient)
  are treated as independent. (25 49 74 pseudo
  patients)
• II. Sample collected from the same patient are
  correlated. (18 42 60 patients)

5
Two groups comparison
6
Data

• Data
• 4CL_ABC_rawIntensity (150168 318 spec, 147
  peaks).
• 4SH_ABC_rawIntensity (150168 318 spec, 88
  peaks).
• 6CL_ABC_rawIntensity (150168 318 spec, 110
  peaks).
• 6SH_ABC_rawIntensity (150168 318 spec, 76
  peaks).
• 9CL_ABC_rawIntensity (150168 318 spec, 154
  peaks).
• 9SH_ABC_rawIntensity (150168 318 spec, 78
  peaks).
• Ciphergene was used to do the pre-process step.
• Each row is one spectra.
• Each column is one peak cluster.
• Each cell is the intensity value.

7
Analysis I linear mixed-effects model

• 7 patients have 3 groups data (sample triplet).
• Goals Two biological meaningful situations
• P_AB P_AC are significant and P_BC are
  insignificant.

8
Analysis I linear mixed-effects model

• Disadvantage
• For each patient, the rejection and
  post-rejection sample are always profiled on the
  same run, but the non-rejection sample is
  profiled on different run. Hence the group C
  effect has been partially confound by the run
  effect.
• Samples collected from the same patient are
  treated as correlated.

9
Analysis I linear mixed-effects model

• Linear mixed-effects model

10
Analysis I linear mixed-effects model

• Linear mixed-effects model

11
Analysis I linear mixed-effects model

• Hypothesis Testing

12
Analysis I linear mixed-effects model

• Results p.ABlt0.05 p.AClt0.05 p.BCgt0.05

13
Analysis I linear mixed-effects model

• Results
• Summary_pAB.txt
• 4CL_ABC_log_plot.ps
• 9SH_ABC_log_plot.ps

14
Analysis II pair-wise comparisons

• No good idea to put all data into one model.
• Pair-wise comparisons.
• A vs B, individual mean. gtp_AB
• A vs C, population mean. gtp_AC
• B vs C, population mean. gtp_BC
• A vs C and B vs C.
• These 7 patients who show up in all three groups
  will be deleted from group C.
• Compare the population mean.
• Relatively less significant.
• Two different assumptions.
• I. Sample collected on different time or profiled
  on different chips (even from the same patient)
  are treated as independent. (254974
  pseudo-patients)
• II. Sample collected from the same patient are
  correlated. (1842 60 individual patients)

15
(No Transcript)
16
Analysis II pair-wise comparisons

• Results.
• Summary_pAB_ind_pairs.txt
• 4CL_ABC_plot_ind_pairs.ps
• 9SH_ABC_plot_ind_pairs.ps
• Summary_pAB_ind_patients.txt
• 4CL_ABC_plot_ind_patients.ps
• 9SH_ABC_plot_ind_patients.ps

17
New Data

• Data
• 4CL_rawIntensity(150168 318 spec, 244 peaks).
• 9SH_rawIntensity (150168 318 spec, ? peaks).
• Pre-process by my own method.
• Each row is one spectra.
• Each column is one peak cluster.
• Each cell is the intensity value.

18
Analysis I linear mixed-effects model

• Results
• Summary_7_patients.txt.
• pH4_CHCA_LOW_plot_7_patients.ps.
• Sig_Peaks_for_peak_2234.665.ps.
• Sig_Peaks_for_peak_4186.782.ps.
• Sig_Peaks_for_peak_5189.455.ps.

19
Analysis II pair-wise comparisons

• Results
• Summary_pairwise_ind_pairs.txt.
• pH4_CHCA_LOW_plot_pairwise_ind_pairs.ps
• Summary_pairwise_ind_patients.txt.
• pH4_CHCA_LOW_plot_pairwise_ind_patients.ps

20

• Thanks!

21
(No Transcript)
22
(No Transcript)
23
(No Transcript)
24
(No Transcript)