Back

The testing of the DAC algorithm for classifying Lymphoma Cell of Origin samples using HTG's targeted gene expression system

Davies, J. R.; Care, M. A.; Mell, T.; Barrans, S.; Burton, C.; Westhead, D. R.

2022-04-18 oncology
10.1101/2022.04.14.22270303 medRxiv
Show abstract

An implementation of the windows based DLBCL automatic classification (DAC) algorithm for determining cell of origin (COO) written in the R language and designed for use with the HTG EdgeSeq Reveal Software package is described. Classifications using the new implementation (DAC-R) were compared to the those using the original version (DAC-Win), the HTG DLBCL COO classifier, the HTG EdgeSeq Reveal DLBCL algorithm. Classifications made using HTG data were tested for concordance with those made using WG-DASL data for the same samples. To analyse small numbers of samples a background dataset was developed to allow for the reliable classification of individual cases. Classification accuracy was assessed using percentage of agreement of discrete classification groups and Pearson correlation of probability scores where appropriate. In the data tested it was seen that the correlation of probability scores for DAC-R and DAC-Win was 0.9985 or higher. Agreement with the HTG DLBCL COO classifier was 85.1% and agreement with the HTG EdgeSeq Reveal DLBCL algorithm was also high (Pearson correlation of GCB probabilities = 0.91). Agreement between classifications made using HTG and WG-DASL data was also high (83.7%). In summary, the R based algorithm of DAC successfully replicates the functionality of the original routine and produces comparable COO calls in data produced from the HTG Pan B-Cell Lymphoma Panel to the other methods listed.

Matching journals

The top 7 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.