What are the requirements of the input?

• All uploaded files are formatted as CSVs.
• Gene expressions need to be raw counts.
• We use hg19 for the annotation.
• Gene expressions [n genes X m individuals]:
  • The size of input is (n+1) rows x (m+1) columns, which including the header as a first row and gene names as a first column.
  • The header [length: m + 1] contains gene following by m individual IDs.
  • n following rows: each row contains ENSEMBL gene name and gene expression of m individuals.
• Mutations [k mutations]:
  • The size of input is (k+1) rows x 2 columns, which including the header as a first row.
  • The input has two columns including ID and GENE.
  • All mutation ids need to be exist in the individual IDs of gene expressions.
  • Gene names are formatted as HUGO symbol.
• Fusions: Same requirements like for mutations.

How can I check the format of input?

Answer: You can download a toy dataset from here. It contains gene expressions of 5 patients from TCGA.

What does “Confidence score” mean?

Answer: Confidence score (CS) is a measurement for how likely “good” IC50 response predictions (IC50 < 1) are confident. CS ranges from 0 to 1. Our recommendation is that the lower threshold of CS should be 0.8

How can I understand some information of an individual?

Answer:

• Subtype prediction: We built a computational model to predict AML subtype for each individual. There are 11 specific subtypes which are collected from Mou et. al

• High confident drugs: Based on the CS threshold you choose, this table give you the brief description of confident drugs.

• Active pathways: Pathway activation score (PAS) > 0. PAS is defined as a tumor-specific pathway activity level that is relevant to a specific drug. It is calculated based on the connection between the drug-specific target gene(s) and the tumor-specific mRNA expression level of the genes in the pathway.

NOTE: MDREAM is for research use only, please do not use results of MDREAM for any practical medical treatment decisions.