IndelRealigner-parameters vs Abra2-parameters

[Manuel-DominguezCBG]

I am comparing IndelRealigner with abra2 (happy to provide my result here later) but first I need to understand and try to match parameters in both tools to do a fair comparison.

My IndelRealigner command is like this

apptainer exec --bind "$ref_folder":"$ref_folder" \
                      "$gatk3" java -Xmx24G -jar /opt/GenomeAnalysisTK.jar \
                      -T IndelRealigner \
                      -targetIntervals demo_data/abra_demo.intervals \
                      -R "$GRCh37" \
                      --maxConsensuses 750 \
                      --maxReadsForConsensuses 3000 \
                      --maxReadsForRealignment 500000 \
                      --maxReadsInMemory 3750000 \
                      -I demo_data/abra_demo.bam \
                      -o demo_data/abra_demo_IR.bam

and my Abra2 command is like this so far

apptainer run --bind "$ref_folder":"$ref_folder" \
                     ./abra2_latest.sif  java -Xmx8G -jar /opt/bin/abra2-2.20.jar \
                        --in demo_data/abra_demo.bam \
                        --out demo_data/abra_demo_ABRA.bam \
                        --ref "$GRCh37" \
                        --threads 8 \
                        --targets demo_data/demo.bed \
                        --mcr 3750000 \
                        --mrr 500000 \

What are the equivalent parameters for maxConsensuses and maxReadsForConsensuses in Abra2??
Does make sense for you to compare mcr with maxReadsInMemory and mrr maxReadsForRealignment?

So far, with the command shown above, the indel in your demo data is realigned with Abra2 but IndelRealigner misses this one.
Abra2

IR

[Manuel-DominguezCBG]

I have another question. One important element I like in IR is that it tells you the indels that were not realigned due to for example excessive number of reads. For us, this is very important becaouse we need to ensure all indels are realignment. Looking at the INFO data in the log file created from ABRA, I cannot see this.