Skip to main content

Sample Data

info

The datasets used for testing and benchmarking MSPC, and benchmarking results are available from Open Science Framework (OSF) at the following link.

https://osf.io/jqrwu/

We use data publicly available from ENCODE to test and benchmark MSPC. This page outlines the specific experiments, our peak calling steps, and links to download the peaks we called and used for testing and benchmarking MSPC.

Dataset v2​

We benchmark MSPC v5 using a cohort containing 48 randomly selected experiments from ENCODE. We call peaks on the samples in each experiment using MACS2 with a permissive threshold as the following --auto-bimodal -p 0.0001 -g hs. This threshold will result in a decreased number of false negatives, with the penalty of an increased number of false positives. We will then reduce the number of false positives while keeping a low rate of false negatives, leveraging combined statistical evidence from replicates (see the methods page). We use this cohort for testing MSPC v5.

The peaks we called on this cohort are available from the following page:

https://osf.io/jqrwu/

Dataset v1​

We benchmarked the first version of MSPC using the dataset v1, which contains 7 experiments selected from ENCODE. We called peaks on this dataset using MACS2 with --auto-bimodal -p 0.0001 -g hs, and the called peaks are available from the following page:

https://osf.io/jqrwu/

The following is the list of the BAM files of the samples in this dataset.