Parameters to fine-tuning iLSI and Harmony process

[wilfridricher]

Hi,

Thanks a lot for this really efficient software with well detailed manual.

We are currently analyzing T cells from different tissues and patients using Multiome (10x Genomics) and we are facing some problems in the clustering probably due to the overintegration of the samples.

To correct the batch effect we used iLSI (using different parameters such as number of iterations, variable nfeatures, clusterParams, etc...) and Harmony.

We obtained a good merging of the different samples, but each time we obtained a large blob of cells without clear separation (see attached picture of the UMAP obtained here) and it is complicated further identify distinct clusters and vizualize difference between gene score expression.

We are wondering if there is a way to improve the process ? Are there ways to optimize iLSI integration ? Or are there other parameters in Harmony that we could use to improve/reduce batch effect correction ?

Thanks a lot in advance!