At first I ran a real-case sample with genome size ~2.6Gb and then instagraal crashed on a 400Gig machine. So I decreased the size of fragments contacts, which was generated from hicstuff (I used several replicates of Hi-C read files to generated contacts information at first. After the machine crashed, I only used one replicate to generate smaller contact files). But still, the smaller sample needs ~30Gig Memory on the host, and 2Gig on the GPU, with 780M abs_fragments_contacts_weighted.txt.
In addition, the program ran super slowly that had ran over 1 month before killed. Before ends, it had generated new fasta file in cycles, taking approximately a day per cycle. But that's only a test sample, so I terminated it in cycle 28.
Nevertheless, could you provide any information about the computational resource usage of the program. Like what I have demonstrated, the 'small' sample seems to run endless... I doubt the possibility to apply it on my real sample.
At first I ran a real-case sample with genome size ~2.6Gb and then
instagraalcrashed on a 400Gig machine. So I decreased the size of fragments contacts, which was generated fromhicstuff(I used several replicates of Hi-C read files to generated contacts information at first. After the machine crashed, I only used one replicate to generate smaller contact files). But still, the smaller sample needs ~30Gig Memory on the host, and 2Gig on the GPU, with 780Mabs_fragments_contacts_weighted.txt.In addition, the program ran super slowly that had ran over 1 month before killed. Before ends, it had generated new fasta file in cycles, taking approximately a day per cycle. But that's only a test sample, so I terminated it in cycle 28.
Nevertheless, could you provide any information about the computational resource usage of the program. Like what I have demonstrated, the 'small' sample seems to run endless... I doubt the possibility to apply it on my real sample.