Question
Question Posted 12/04/17:
Hello,
I was using the UPENNBIOMK_MASTER table and noticed there are some rows with the same RID and VISCODE, but different values of abeta, tau and ptau. Those rows have different batch, why does that happen? Also, the rows with batch "MEDIAN" have the abeta, tau, ptau filled but abeta_raw, tau_raw and ptau_raw are empty, shouldn't that be the contrary?
These are some rows that demonstrate the problem:
RID VISCODE BATCH KIT STDS DRWDTE RUNDATE ABETA TAU PTAU ABETA_RAW TAU_RAW PTAU_RAW update_stamp
243 bl UPENNBIOMK2 190841 181286 2006-04-13 2008-10-31 141.0 160.0 55.6 161.0 162.0 34.5 2016-07-06 16:15:51.0
243 bl MEDIAN ALL ALL 2006-04-13 2016-03-09 144.0 155.0 52.8 NaN NaN NaN 2016-07-06 16:15:51.0
256 bl UPENNBIOMK2 190841 181286 2006-04-20 2008-11-06 84.7 94.7 42.8 86.9 96.1 26.8 2016-07-06 16:15:51.0
256 bl UPENNBIOMK3 D191113 191075 2006-04-20 2009-08-12 113.0 111.0 27.0 51.9 40.0 25.0 2016-07-06 16:15:51.0
Hello,
I was using the UPENNBIOMK_MASTER table and noticed there are some rows with the same RID and VISCODE, but different values of abeta, tau and ptau. Those rows have different batch, why does that happen? Also, the rows with batch "MEDIAN" have the abeta, tau, ptau filled but abeta_raw, tau_raw and ptau_raw are empty, shouldn't that be the contrary?
These are some rows that demonstrate the problem:
RID VISCODE BATCH KIT STDS DRWDTE RUNDATE ABETA TAU PTAU ABETA_RAW TAU_RAW PTAU_RAW update_stamp
243 bl UPENNBIOMK2 190841 181286 2006-04-13 2008-10-31 141.0 160.0 55.6 161.0 162.0 34.5 2016-07-06 16:15:51.0
243 bl MEDIAN ALL ALL 2006-04-13 2016-03-09 144.0 155.0 52.8 NaN NaN NaN 2016-07-06 16:15:51.0
256 bl UPENNBIOMK2 190841 181286 2006-04-20 2008-11-06 84.7 94.7 42.8 86.9 96.1 26.8 2016-07-06 16:15:51.0
256 bl UPENNBIOMK3 D191113 191075 2006-04-20 2009-08-12 113.0 111.0 27.0 51.9 40.0 25.0 2016-07-06 16:15:51.0
Response posted 12/04/17 by Michal Figurski:
Hello,
Just to answer your question shortly: due to lot-to-lot variability of subsequent reagent lots used to process CSF samples, ADNI has implemented a rule for longitudinal samples, that the samples from all visits of a particular patient should be run on the same plate, with a single reagent lot. As with time the new longitudinal samples were added for many patients, the growing longitudinal sets were processed multiple times, resulting in multiple results for the same patient/visit. These results are reported in batches named UPENNBIOMKx.csv - where 'x' is the batch number.
Secondly, because these results from subsequent batches were at times in a different scale (lot to lot reagents differences), they were brought to common scale by the process of 'anchoring' to the first batch. Hence each subsequent batch has 'raw' results - the original results before anchoring.
The UPENNBIOMK_MASTER combines multiple batches of results into a single file to make them conveniently accessible in one place. In addition to batched results, it also provides calculated medians - a median is obtained for a single patient/visit, from all available batches. Medians only make sense for results in common scale (anchored), and does not make sense for raw data.
For more details on this - please look into the 'methods summary' documents on LONI, for example this one: 'ADNI_Methods_Shaw_Trojanowski_Corrected_Anchoring_of_2012_dataset_31-Oct-2013.pdf', and also others."
Just to answer your question shortly: due to lot-to-lot variability of subsequent reagent lots used to process CSF samples, ADNI has implemented a rule for longitudinal samples, that the samples from all visits of a particular patient should be run on the same plate, with a single reagent lot. As with time the new longitudinal samples were added for many patients, the growing longitudinal sets were processed multiple times, resulting in multiple results for the same patient/visit. These results are reported in batches named UPENNBIOMKx.csv - where 'x' is the batch number.
Secondly, because these results from subsequent batches were at times in a different scale (lot to lot reagents differences), they were brought to common scale by the process of 'anchoring' to the first batch. Hence each subsequent batch has 'raw' results - the original results before anchoring.
The UPENNBIOMK_MASTER combines multiple batches of results into a single file to make them conveniently accessible in one place. In addition to batched results, it also provides calculated medians - a median is obtained for a single patient/visit, from all available batches. Medians only make sense for results in common scale (anchored), and does not make sense for raw data.
For more details on this - please look into the 'methods summary' documents on LONI, for example this one: 'ADNI_Methods_Shaw_Trojanowski_Corrected_Anchoring_of_2012_dataset_31-Oct-2013.pdf', and also others."
Response posted 12/05/17 by Michal Figurski:
Hello,
Just to answer your question shortly: due to lot-to-lot variability of subsequent reagent lots used to process CSF samples, ADNI has implemented a rule for longitudinal samples, that the samples from all visits of a particular patient should be run on the same plate, with a single reagent lot. As with time the new longitudinal samples were added for many patients, the growing longitudinal sets were processed multiple times, resulting in multiple results for the same patient/visit. These results are reported in batches named UPENNBIOMKx.csv - where 'x' is the batch number.
Secondly, because these results from subsequent batches were at times in a different scale (lot to lot reagents differences), they were brought to common scale by the process of 'anchoring' to the first batch. Hence each subsequent batch has 'raw' results - the original results before anchoring.
The UPENNBIOMK_MASTER combines multiple batches of results into a single file to make them conveniently accessible in one place. In addition to batched results, it also provides calculated medians - a median is obtained for a single patient/visit, from all available batches. Medians only make sense for results in common scale (anchored), and does not make sense for raw data.
For more details on this - please look into the 'methods summary' documents on LONI, for example this one: 'ADNI_Methods_Shaw_Trojanowski_Corrected_Anchoring_of_2012_dataset_31-Oct-2013.pdf', and also others."
Just to answer your question shortly: due to lot-to-lot variability of subsequent reagent lots used to process CSF samples, ADNI has implemented a rule for longitudinal samples, that the samples from all visits of a particular patient should be run on the same plate, with a single reagent lot. As with time the new longitudinal samples were added for many patients, the growing longitudinal sets were processed multiple times, resulting in multiple results for the same patient/visit. These results are reported in batches named UPENNBIOMKx.csv - where 'x' is the batch number.
Secondly, because these results from subsequent batches were at times in a different scale (lot to lot reagents differences), they were brought to common scale by the process of 'anchoring' to the first batch. Hence each subsequent batch has 'raw' results - the original results before anchoring.
The UPENNBIOMK_MASTER combines multiple batches of results into a single file to make them conveniently accessible in one place. In addition to batched results, it also provides calculated medians - a median is obtained for a single patient/visit, from all available batches. Medians only make sense for results in common scale (anchored), and does not make sense for raw data.
For more details on this - please look into the 'methods summary' documents on LONI, for example this one: 'ADNI_Methods_Shaw_Trojanowski_Corrected_Anchoring_of_2012_dataset_31-Oct-2013.pdf', and also others."
