SMLG (Statistical Machine Learning Group) Discussion Forum

Using this master file, we will perform various case control analyses, such as:

1. Normal-8 samples to basal-71 samples
2. 8 normal to 53 her2 rich
3. 8 normal to 217 luminal A
4. 8 normal to 111 luminal B
5. 71 basal to 217 luminal A
6. 71 basal to 111 luminal B
7. 71 basal to 53 HER-enriched
8. 463 cases (samples) where we combine all subtypes into a variable called PAM50_Subtype

cwyoo wrote:Using this master file, we perform various case control analyses, such as:

1. Normal-8 samples to basal-71 samples
2. 8 normal to 53 her2 rich
3. 8 normal to 217 luminal A
4. 8 normal to 111 luminal B
5. 71 basal to 217 luminal A
6. 71 basal to 111 luminal B
7. 71 basal to 53 HER-enriched
8. 463 cases (samples) where we combine all subtypes into a variable called PAM50_Subtype

This is a dataset that was created by identifying 933 genes that are significantly correlated with NRF1 (r > 0.09191, p < 0.05, see my posting under: Board index ‹ Analyses & Results ‹ Others ‹ Breast Cancer ‹ Datasets, posted on Tue Mar 03, 2015 8:14 pm (http://131.94.9.17/phpbb/viewtopic.php?f=49&t=50&start=40#p359) how we calculated the p value.

Also this dataset introduces PAM50_Subtype which codes the following number of subtypes: 8 Normal-like (coded as 0), 81 Basal-like (coded as 1), 209 Luminal A (coded as 2), 112 Luminal B (coded as 3), and 53 HER2-enriched (coded as 4)

cwyoo wrote:

cwyoo wrote:Using this master file, we perform various case control analyses, such as:

1. Normal-8 samples to basal-71 samples
2. 8 normal to 53 her2 rich
3. 8 normal to 217 luminal A
4. 8 normal to 111 luminal B
5. 71 basal to 217 luminal A
6. 71 basal to 111 luminal B
7. 71 basal to 53 HER-enriched
8. 463 cases (samples) where we combine all subtypes into a variable called PAM50_Subtype

This is a dataset that was created by identifying 933 genes that are significantly correlated with NRF1 (r > 0.09191, p < 0.05, see my posting under: Board index ‹ Analyses & Results ‹ Others ‹ Breast Cancer ‹ Datasets, posted on Tue Mar 03, 2015 8:14 pm (http://131.94.9.17/phpbb/viewtopic.php?f=49&t=50&start=40#p359) how we calculated the p value.

Also this dataset introduces PAM50_Subtype which codes the following number of subtypes: 8 Normal-like (coded as 0), 81 Basal-like (coded as 1), 209 Luminal A (coded as 2), 112 Luminal B (coded as 3), and 53 HER2-enriched (coded as 4)

Due to limited memory, banjo could not handle 935 variables. So we selected genes that are significantly correlated with NRF1 with p < 0.01, i.e., |r| > 0.12135. Total of 655 genes were selected. Also, due to limited number of cases of Normal-like, we will only perform the following analyses:

1. 81 basal to 209 luminal A
2. 81 basal to 112 luminal B
3. 81 basal to 53 HER-enriched
4. 463 cases (samples) where we combine all subtypes into a variable called PAM50_Subtype

Sounds good idea! Where is the data with only 655 genes for banjo analysis>

droy wrote:Sounds good idea! Where is the data with only 655 genes for banjo analysis>

Here they are.

cwyoo wrote:

droy wrote:Sounds good idea! Where is the data with only 655 genes for banjo analysis>

Here they are.

36 hour banjo run result of Basal vs Lumina A. This network was 1.64 e36 times better than the second best network.

I do not see Lum A in the network.

droy wrote:I do not see Lum A in the network.

It is highlighted and labeled "Basal_LumA" and the same label is used in the dataset.

Dr. Yoo,

As spoken earlier, you wanted me to run Banjo on the following files for 24 hours:

3. 81 basal to 53 HER-enriched
4. 463 cases (samples) where we combine all subtypes into a variable called PAM50_Subtype

Nevertheless, I'm unclear which one you want me to run Bene for? Is it for both these files as well?

meninonas wrote:Dr. Yoo,

As spoken earlier, you wanted me to run Banjo on the following files for 24 hours:

3. 81 basal to 53 HER-enriched
4. 463 cases (samples) where we combine all subtypes into a variable called PAM50_Subtype

Nevertheless, I'm unclear which one you want me to run Bene for? Is it for both these files as well?

Yes. Run Bene on all four datasets.