Hello. When I modify the complexIndels setting to be TRUE -- as opposed to the default setting of false -- I have a shorter runtime and actually find slightly fewer indels. I am confused by this; I would have thought setting this to TRUE would result in longer times and more indels. I started looking into this, after a similar finding that setting indel length to 1 (indels=1 instead of default of 5) dramatically reduced run time. Can you help me understand these settings? My purpose in the analysis is differential gene expression based on total RNA-Seq, so indels and variants are not a key factor - but I am trying to ensure I optimize settings for efficiency with large datasets, and to obtain the best mapping results. Thank you in advance for your help!

#CODE WITH COMPLEX INDELS AS TRUE

align(index=MyIndex, readfile1=R1, readfile2=R2, type="rna", minFragLength=50, maxFragLength=600, maxMismatches=3, unique=TRUE, nthreads=7, output_format="BAM", indels=1, TH1=3, TH2=1, complexIndels=TRUE);

#Results as follows:
#Total_fragments                                                                          81428563
#Mapped_fragments                                                                         73728794
#Uniquely_mapped_fragments                                                                73728794
#Multi_mapping_fragments                                                                         0
#Unmapped_fragments                                                                        7699769
#Properly_paired_fragments                                                                69079933
#Singleton_fragments                                                                        743665
#More_than_one_chr_fragments                                                                 78673
#Unexpected_strandness_fragments                                                              7749
#Unexpected_template_length                                                                2547646
#Inversed_mapping                                                                          1271128
#Indels                                                                                     543377
#Running time : 59.5 minutes

#CODE WITH DEFAULT, SO COMPLEX INDELS IS FALSE

align(index=MyIndex, readfile1=R1, readfile2=R2, type="rna", minFragLength=50, maxFragLength=600, maxMismatches=3, unique=TRUE, nthreads=7, output_format="BAM", indels=1, TH1=3, TH2=1);


#Results as follows:
#Total_fragments                                                                          81428563
#Mapped_fragments                                                                         73725382
#Uniquely_mapped_fragments                                                                73725382
#Multi_mapping_fragments                                                                         0
#Unmapped_fragments                                                                        7703181
#Properly_paired_fragments                                                                69067023
#Singleton_fragments                                                                        703253
#More_than_one_chr_fragments                                                                 77453
#Unexpected_strandness_fragments                                                              8606
#Unexpected_template_length                                                                2543287
#Inversed_mapping                                                                          1325760
#Indels                                                                                     544768
#Running time : 99.1 minutes

sessionInfo()
R version 4.0.3 (2020-10-10)
Platform: x86_64-redhat-linux-gnu (64-bit)
Running under: Red Hat Enterprise Linux 8.3 (Ootpa)

Matrix products: default
BLAS/LAPACK: /usr/lib64/libopenblas-r0.3.3.so

locale:
 [1] LC_CTYPE=en_US.UTF-8       LC_NUMERIC=C              
 [3] LC_TIME=en_US.UTF-8        LC_COLLATE=en_US.UTF-8    
 [5] LC_MONETARY=en_US.UTF-8    LC_MESSAGES=en_US.UTF-8   
 [7] LC_PAPER=en_US.UTF-8       LC_NAME=C                 
 [9] LC_ADDRESS=C               LC_TELEPHONE=C            
[11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C       

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
[1] Rsubread_2.4.2

loaded via a namespace (and not attached):
[1] compiler_4.0.3  Matrix_1.3-2    grid_4.0.3      lattice_0.20-41

We investigated this but we could not reproduce what you found regarding the use of the complexIndels parameter. We did not find setting complexIndels=TRUE results in less running time. The running time remains pretty much unchanged in our testing.

When complexIndels is set to TRUE, the Subread aligner is tuned to put more focus on detecting complex indels which are short indels that are located very close to each other. This results in less 'normal' indels being detected and this explains why less indels were reported for your data when you set complexIndels=TRUE.

I would not recommend setting complexIndels=TRUE when your purpose is to perform differential expression analysis of your RNA-seq data. Detecting complex indels is unlikely to be helpful for your expression analysis because complex indels are rare and detecting such indels weakens Subreads ability to detect normal indels.

Although your evaluation results seem to suggest that the mapping and counting results are quite similar between the setting indels=1 and setting indels=5, I am still reluctant to use the setting indels=1 to try to speed up the read mapping because you will have an increased number of incorrect mappings. This will be particularly problematic for samples that have a lot of mutations such as cancer samples.

I strongly recommend you using the default setting as much as possible as these settings have been well tested and they were found to work well in most circumstances from our experiences.