Page Comparison

{color:#ff0000}{_}Important update (January 20th, 2011): the data below have been corrected for the BCR batch which is not necessarily the processing batch. The dataset needs to be reanalyzed._{color}_ _

_Batch on the download page,"# after ""HumanMethylation27k"" in the file name, Level 1",Batch as the sixth field in the patient barcode,Comments_
Batch 34,3,0752,
Batch 37,"1, 2","0945, 1104","0945 comes from 1, one sample that is 1104 comes from 2"
Batch 52,2,1104,
Batch 58,4,1205,
Batch 84,no data,,
Batch 119,no data,,
Batch 144,no data,,
Batch 160,no data,,
Batch 166,no data,,
Batch 183,no data,,

h4. Correlation between BCR batch and the processing batch for 27k arrays (January 20, 2012)
{csv}Batch on the download page,"# after ""HumanMethylation27k"" in the file name, Level 1",Batch as the sixth field in the patient barcode
Batch 47,1,A00Y
Batch 56,2,A032
Batch 61,no data,
Batch 72,no data,
Batch 74,no data,
Batch 80,no data,
Batch 85,3,A112
Batch 93,4,A12E
Batch 96,no data,
Batch 103,no data,
Batch 109,no data,
Batch 117,no data,
Batch 120,no data,
Batch 124,no data,
Batch 136,no data,
Batch 142,no data,
Batch 147,no data,
Batch 155,no data,
Batch 167,no data,
Batch 177,no data,
Batch 185,no data,{csv}

h5. Analysis of batch vs clinical traits

Number of clinical traits: 84

Number of batches based on tumor DNA methylation data (samples retrieved according to this pattern: "TCGA-..-....-0..-..D-....-05"): 24

Correlation between center and batches ('two'=center (second field in the patient barcode)):
{code:collapse=true}> table(batchID,two)
       two
batchID A1 A2 A7 A8 AC AN AO AQ AR B6 BH C8 D8 E2 E9 EW GI GM HN
   A00Y  0  3  7 66  0 12  2  0  0  0  4  0  0  0  0  0  0  0  0
   A032  0 22  0 14  0 19 10  1  0 16 10  0  0  0  0  0  0  0  0
   A058  0  2  1  3  0  0 11  0  0  4 26  0  0  0  0  0  0  0  0
   A088  7 14  0  1  0  0  1  0  8  9  7  0  0  0  0  0  0  0  0
   A10A  0 12  0  0  0  9  0  0  5  9  4  0  0  0  0  0  0  0  0
   A10N  0  1  0  0  0  1 12  2  0  1  1  0  0  9  0  0  0  0  0
   A10P  7 16  1  4  0  0 12  0  8 13 32  0  0  0  0  0  0  0  0
   A112  1 11  1  0  0  0  5  0  3  5 18 20  9 15  0  0  0  0  0
   A12E  0  0  0  0  0  0  0  0  1  0 20  2  0 21  0  0  0  0  0
   A12R  0  0  3  0  0  0  0  0 15  0 14  0  0 11  0  0  0  0  0
   A138  0  0  0  0  0  0  0  0  1  0  7  6  0  2  0  0  0  0  0
   A13K  0  7  1  0  0  0  5  2  0  3 18  4 19  6  0  8  0  0  0
   A145  6  0  0  0  0  0  1  0  0  0  0  0  0 10  4 19  0  0  0
   A148  0  0  1  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0
   A14H  0  1  0  0  0  0  0  0  0  0  1  0  7  5 10  1  0  0  0
   A14N  0  0  0  0  0  0  0  1  0  1  2  0 20  1  6  1  0  0  0
   A161  0  0  0  0  2  0  0  0  0  1  3  0  3  2 17  0  0  0  0
   A16A  0  4  6  0  1  0  0  0 22  0  1  3  0  0  9  0  0  0  0
   A16G  0  3  0  0  0  0  0  0  0  0  1  8 13  0  3  0  1  0  0
   A17F  0  0  0  0  4  0  0  0  0  0  0  0  1  0  0  3  0  0  0
   A17Z  0  0  0  0  2  0  0  0  4  0  0  0  0  0  1  0  0  6  0
   A18O  0  0  0  0  1  0  0  0  5  1  1  0  0  0  1  0  0  7  1
   A19F  0  2  0  1  0  0  0  0  0  1  0  0  0  1  0  0  0  0  0
   A19Z  0  0  0  0  5  0  0  0  1  0  1  0  0  1  0  0  0  0  0
{code}

Significant batch-clinical traits correlations (the entire list can be found [here|^BatchClinicalInfoCorrelationsBRCA.csv]):

{csv}
"BRCA_clinical_traits","DataType","NumberOfNAs","Test","Pvalue"
"tissue_prospective_collection_indicator","factor",35,"Pearson's Chi-squared test",4.47E-62
"tissue_retrospective_collection_indicator","factor",35,"Pearson's Chi-squared test",4.47E-62
"year_of_initial_pathologic_diagnosis","integer",34,"Kruskal-Wallis rank sum test",3.15E-32
"breast_carcinoma_first_surgical_procedure_name","factor",54,"Pearson's Chi-squared test",5.45E-32
"days_to_last_followup","integer",73,"Kruskal-Wallis rank sum test",3.07E-31
"days_to_form_completion","integer",34,"Kruskal-Wallis rank sum test",5.70E-31
"first_pathologic_diagnosis_biospecimen_acquisition_method_type","factor",123,"Pearson's Chi-squared test",3.39E-28
"breast_tumor_clinical_m_stage","factor",35,"Pearson's Chi-squared test",1.06E-22
"axillary_lymph_node_stage_method_type","factor",223,"Pearson's Chi-squared test",9.33E-19
"breast_tumor_pathologic_n_stage","factor",34,"Pearson's Chi-squared test",2.19E-17
"lab_proc_her2_neu_immunohistochemistry_receptor_status","factor",41,"Pearson's Chi-squared test",6.22E-16
"breast_carcinoma_estrogen_receptor_status","factor",34,"Pearson's Chi-squared test",1.85E-13
"breast_carcinoma_progesterone_receptor_status","factor",34,"Pearson's Chi-squared test",8.87E-13
"vital_status","factor",34,"Pearson's Chi-squared test",2.38E-09
"anatomic_site_location_descriptor","factor",119,"Pearson's Chi-squared test",1.03E-07
"age_at_initial_pathologic_diagnosis","integer",34,"Kruskal-Wallis rank sum test",5.87E-06
"days_to_birth","integer",34,"Kruskal-Wallis rank sum test",6.68E-06
"lab_procedure_her2_neu_in_situ_hybrid_outcome_type","factor",194,"Pearson's Chi-squared test",3.18E-05
"person_menopause_status","factor",161,"Pearson's Chi-squared test",5.70E-05
"breast_tumor_pathologic_grouping_stage","factor",40,"Pearson's Chi-squared test",7.40E-05
"her2_immunohistochemistry_level_result","factor",351,"Pearson's Chi-squared test",1.72E-04
"breast_tumor_pathologic_t_stage","factor",34,"Pearson's Chi-squared test",2.82E-04
"pos_finding_lymph_node_hematoxylin_and_eosin_staining_microscopy_count","integer",177,"Kruskal-Wallis rank sum test",6.49E-04
"cytokeratin_immunohistochemistry_staining_method_micrometastasis_indicator","factor",324,"Pearson's Chi-squared test",8.61E-04
"person_neoplasm_cancer_status","factor",284,"Pearson's Chi-squared test",7.95E-03
"breast_cancer_optical_measurement_histologic_type","factor",34,"Pearson's Chi-squared test",1.47E-02
"disease_surgical_margin_status","factor",82,"Pearson's Chi-squared test",3.70E-02
{csv}

h5. Correlation with survival

Relevant clinical traits: days to the last follow-up (27), vital status (83), days to death (24), days to last know alive (28), summaries:
{code:collapse=true}> summary(clinical[,27]) # days to the last follow up
   Min. 1st Qu.  Median    Mean 3rd Qu.    Max.    NA's
    0.0   140.0   457.0   815.8  1194.0  6795.0    73.0
> table(clinical[,83]) #vital status

DECEASED   LIVING
      93      725
> summary(clinical[,24]) # days to death
   Min. 1st Qu.  Median    Mean 3rd Qu.    Max.    NA's
    157     811    1563    1744    2520    4456     759
> summary(clinical[,28]) # days to last known alive
   Min. 1st Qu.  Median    Mean 3rd Qu.    Max.    NA's
    0.0   293.5   607.5  1068.0  1442.0  6795.0   508.0{code}
It seems that similarly to the colon cancer combined datasets days to last known alive is similar to the days to the last follow-up, however days to the last follow up contains more information (fewer NAs), use it for construction of the survival object. No patients missed information for both days to the last follow up and days to death. The survival object was created in the same way as for the analyses of other TCGA cancer datasets. Info is available ([here|METHYLATION:Colon cancer] and [here|METHYLATION:AML - acute myeloid leukemia]) 

Kaplan Meier curve and survival plots break down by batch:
!BRCA_KaplanMeierCurve.png|thumbnail! !BRCA_SurvivalVsBatch.png|thumbnail!
Here is the summary of the survival vs batch:
{code:collapse=true}
> summary(coxph(surv~methM[,2]))
Call:
coxph(formula = surv ~ methM[, 2])

  n= 818, number of events= 93
   (34 observations deleted due to missingness)

                     coef  exp(coef)   se(coef)      z Pr(>|z|)
methM[, 2]A032 -6.961e-01  4.985e-01  5.148e-01 -1.352   0.1763
methM[, 2]A058 -2.321e+00  9.819e-02  1.084e+00 -2.140   0.0323 *
methM[, 2]A088 -5.512e-01  5.763e-01  5.413e-01 -1.018   0.3086
methM[, 2]A10A -2.238e-01  7.995e-01  5.565e-01 -0.402   0.6876
methM[, 2]A10N -1.455e+00  2.334e-01  8.217e-01 -1.771   0.0766 .
methM[, 2]A112 -9.643e-01  3.812e-01  5.847e-01 -1.649   0.0991 .
methM[, 2]A12E  9.108e-01  2.486e+00  5.015e-01  1.816   0.0694 .
methM[, 2]A12R -1.794e+00  1.663e-01  1.082e+00 -1.657   0.0975 .
methM[, 2]A138  8.921e-01  2.440e+00  5.487e-01  1.626   0.1040
methM[, 2]A13K  2.542e-01  1.289e+00  4.825e-01  0.527   0.5983
methM[, 2]A145 -9.748e-01  3.773e-01  1.081e+00 -0.902   0.3673
methM[, 2]A14H  3.164e-01  1.372e+00  8.215e-01  0.385   0.7002
methM[, 2]A14N  9.591e-01  2.609e+00  1.089e+00  0.880   0.3786
methM[, 2]A161 -1.382e-01  8.709e-01  7.183e-01 -0.192   0.8475
methM[, 2]A16A -1.347e+00  2.600e-01  8.206e-01 -1.642   0.1007
methM[, 2]A16G -1.714e+01  3.615e-08  5.030e+03 -0.003   0.9973
methM[, 2]A17F -1.731e+01  3.039e-08  1.338e+04 -0.001   0.9990
methM[, 2]A17Z -1.725e+01  3.216e-08  4.025e+03 -0.004   0.9966
methM[, 2]A18O -1.434e+00  2.383e-01  1.088e+00 -1.318   0.1875
methM[, 2]A19Z         NA         NA  0.000e+00     NA       NA
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

               exp(coef) exp(-coef) lower .95 upper .95
methM[, 2]A032 4.985e-01  2.006e+00   0.18176    1.3672
methM[, 2]A058 9.819e-02  1.018e+01   0.01172    0.8223
methM[, 2]A088 5.763e-01  1.735e+00   0.19946    1.6649
methM[, 2]A10A 7.995e-01  1.251e+00   0.26858    2.3797
methM[, 2]A10N 2.334e-01  4.284e+00   0.04664    1.1685
methM[, 2]A112 3.812e-01  2.623e+00   0.12121    1.1991
methM[, 2]A12E 2.486e+00  4.022e-01   0.93037    6.6440
methM[, 2]A12R 1.663e-01  6.012e+00   0.01993    1.3879
methM[, 2]A138 2.440e+00  4.098e-01   0.83256    7.1522
methM[, 2]A13K 1.289e+00  7.755e-01   0.50082    3.3197
methM[, 2]A145 3.773e-01  2.651e+00   0.04531    3.1409
methM[, 2]A14H 1.372e+00  7.288e-01   0.27424    6.8653
methM[, 2]A14N 2.609e+00  3.832e-01   0.30851   22.0693
methM[, 2]A161 8.709e-01  1.148e+00   0.21309    3.5598
methM[, 2]A16A 2.600e-01  3.846e+00   0.05206    1.2985
methM[, 2]A16G 3.615e-08  2.766e+07   0.00000       Inf
methM[, 2]A17F 3.039e-08  3.290e+07   0.00000       Inf
methM[, 2]A17Z 3.216e-08  3.109e+07   0.00000       Inf
methM[, 2]A18O 2.383e-01  4.196e+00   0.02825    2.0108
methM[, 2]A19Z        NA         NA        NA        NA

Rsquare= 0.069   (max possible= 0.67 )
Likelihood ratio test= 58.61  on 19 df,   p=6.4e-06
Wald test            = 50  on 19 df,   p=0.0001311
Score (logrank) test = 69.46  on 19 df,   p=1.129e-07

Warning messages:
1: In fitter(X, Y, strats, offset, init, control, weights = weights,  :
  Loglik converged before variable  16,17,18 ; beta may be infinite.
2: In coxph(surv ~ methM[, 2]) :
  X matrix deemed to be singular; variable 20{code}
_It seems that there are a lot of errors, I wonder why. I also don't understand where those observations come from that are deleted due to missingness. Need to ask someone to help clarify this output_. Update (January 5, 2012): there are NAs for some batches because I had factor levels left in the batch vector but no data for those levels. Fixed the problem with that. "Deleted due to missingness" also fixed as I figured out how that I need to be more careful about using 'match' for subsetting. 

h5. DNA methylation data

December 21st, 2011: 27k and 450k arrays are available. Downloaded Level 1 450k data. It seems that they started splitting green and red probes into 2 separate files and they also provide now the Illumina's idat files which are the bead level data (not tab delimited files). I need to find a way to process them, it seems that Bioconductor beadarray package can be used to read these files and do some bead level normalization (summarization too?). The Level2 data contains already summarized and normalized data (tab delimited files with CpG ID, value for methylated and value for unmethylated probes), however it is available only for 91 patients. Also tried to download 27k arrays available for breast cancer, however the data is available for \~26 patients (they stopped running those arrays?).  I guess I need to figure out how to process Level 1 data.