HiCCUPS will not run on an interaction matrix that is too sparse (less than 300m filtered reads), and so for our low cell number samples which were not sequenced as deeply, we used the Fit-HiC contact caller. We also processed our 25m cell libraries with Fit-HiC for the comparisons in Figure 1d and Supplementary Figure 2a .
Bin pairs of the interaction matrix with statistically significant contact signal were identified using Fit-HiC19 (link). Genome wide intra-chromosomal bin pairs were filtered for an interaction distance between 20kb and 2Mb, and default Fit-HiC settings were used to calculate false discovery rate (FDR) values for each bin-pair in a given HiChIP experiment. For the comparison of mESC experiments inFigure 1D , Smc1a HiChIP binpairs or Smc1a ChIA-PET high-confidence interactions13 (link) were filtered for overlap with Smc1a ChIP-seq peaks. The appropriate FDR was selected for each HiChIP experiment to result in approximately 10,000 contacts per experiment.
GM12878 Smc1a HiChIP filtered PETs from the HiC-Pro pipeline were processed through stages 4 and 5 of the Mango17 (link) pipeline to call significant interactions. Stage 4 applies MACS218 to call peaks using PETs. Stage 5 modeled the background interactions by taking interaction distance and depth into consideration and used a binomial distribution to call significant interactions. With our Mango parameters of PETs >= 4 and a FDR cutoff of 10−4 we obtained 61395 significant interactions.
The Juicer pipeline’s HiCCUPS tool was used to identify loops5 (link),20 (link),21 (link). Filtered read-pairs from the HiC-Pro pipeline were converted into .hic format files and input into HiCCUPS. The same parameters used for GM12878 Hi-C5 (link),20 (link),21 (link) were used on the HiChIP datasets as follows: hiccups -m 500 -r 5000,10000 -f 0.1,0.1 -p 4,2 -i 7,5 -d 20000,20000 HiCCUPS_output.txt
Bin pairs of the interaction matrix with statistically significant contact signal were identified using Fit-HiC19 (link). Genome wide intra-chromosomal bin pairs were filtered for an interaction distance between 20kb and 2Mb, and default Fit-HiC settings were used to calculate false discovery rate (FDR) values for each bin-pair in a given HiChIP experiment. For the comparison of mESC experiments in
GM12878 Smc1a HiChIP filtered PETs from the HiC-Pro pipeline were processed through stages 4 and 5 of the Mango17 (link) pipeline to call significant interactions. Stage 4 applies MACS218 to call peaks using PETs. Stage 5 modeled the background interactions by taking interaction distance and depth into consideration and used a binomial distribution to call significant interactions. With our Mango parameters of PETs >= 4 and a FDR cutoff of 10−4 we obtained 61395 significant interactions.
The Juicer pipeline’s HiCCUPS tool was used to identify loops5 (link),20 (link),21 (link). Filtered read-pairs from the HiC-Pro pipeline were converted into .hic format files and input into HiCCUPS. The same parameters used for GM12878 Hi-C5 (link),20 (link),21 (link) were used on the HiChIP datasets as follows: hiccups -m 500 -r 5000,10000 -f 0.1,0.1 -p 4,2 -i 7,5 -d 20000,20000 HiCCUPS_output.txt