about discarding clusters

[firatuyulur]

Hi, I want to understand more about how lichee discards the clusters while building a tree. In my setup I am using PyCloneVI for clonal deconvolution. I am providing both the cluster profile file and the snv file.

java -jar lichee.jar -build \
            -n 0 \
            -i {input.muts} \
            -clustersFile {input.clusters} \
            -sampleProfile \
            -o {output.out} \
            -v \
            -cp \
            -maxVAFAbsent 0.005 \
            -minVAFPresent 0.005 \
            -s 10

max and min VAF parameters may become irrelevant the way I run lichee but here is how I run it. I've also attached example input files for a sample. The run will have 3 clusters discarded.

According to the paper due to noise error margin being too narrow or misclassified binary profile, there can be cases with no valid lineage trees. And lichee starts discarding non-robust groups starting from the smallest nodes until it can build a tree. In my setup I am providing PyCloneVI clusters with at least 5 SNVs assigned. What is "non-robust" when a setup like I have is used? is there a way to attempt to get trees discarding random non-robust nodes instead of the smallest first? Would you have a suggestion to reduce the amount of trees discareded?

Thanks very much

for_lichee_from_pyclonevi_nomaxcluster_CP_CRUK0018_beta-binomial_snvs_vaf005.input.txt
for_lichee_from_pyclonevi_nomaxcluster_CP_CRUK0018_beta-binomial_clusters_vaf005.input.txt
.