TheJacksonLaboratory / sbas

CloudOS Digital Research Environment
4 stars 5 forks source link

YARN #154

Closed pnrobinson closed 4 years ago

pnrobinson commented 4 years ago

This is how we describe YARN right now. Can we recycle this description? Anne wrote a new function for YARN if memory serves, should we mention that?

Normalization of counts data for alternative splicing analysis

We used the Yet Another RNA Normalization software pipeline (YARN) to look for samples that are likely to be mis-annotated (Paulson et al., 2017). We applied the function checkMisAnnotation using chromosome Y genes as control genes and removed the individual GTEX-11ILO from the dataset (similar to ref. (Paulson et al., 2017)). We followed the YARN preprocessing procedure for identifying GTEx tissues that can be combined in the differential splicing analysis, using the function checkTissuesToMerge. This function creates multidimensional scaling (MDS) plots that reveal similarities and differences between samples in a set of tissues. As input for MDS, we concatenated the matrices of the skip and inclusion isoform counts row-wise such that each data point in the MDS is the counts of skip and inclusion isoforms in one sample, where a sample corresponds to an individual and tissue. In order to obtain a consistent criterion for merging, we created an MDS for each pair of samples in the following regions: brain, artery, esophagus, skin and fibroblasts, colon and adipose. We calculated the normalized distance () between samples of the same anatomical region defined as the intersample distance divided by the mean distance of all samples for each pairwise-MDS and AS event type. We then merged tissues where0.95 for all AS event types. In skin regions the merging was identical to that in the original YARN publication, and in brain regions our procedure further refined merges performed in YARN into finer subsets. For esophagus, we found that Esophagus - Gastroesophageal Junction and Esophagus - Muscularis can be merged. Similar to YARN, other regions were not found to be mergeable with respect to the 5 types of AS events investigated in our study.

cgpu commented 4 years ago

As far as the version of yarn with @adeslatt 's updates:

The version we are using is this one https://github.com/TheJacksonLaboratory/sbas/blob/c13f2e05dbf86d488bbabb24fa3f1cd215f3e795/dependencies/install.R#L4

The at @3ff72c0 in the installation command here denotes the github revisions of the TheJacksonLaboratory/yarn repo were Anne has applied her updates.

adeslatt commented 4 years ago

I will update these words -- I did end up doing something different this time arround -- I got this!

pnrobinson commented 4 years ago

duplicate with #192