Closed enricorox closed 1 year ago
Ok, the exception was due to blossom5
and not matchtigs
. Maybe a warning about the memory requirement can be helpful since it requires O(|V|^2) if I'm correct.
Hi, thank you for trying out matchtigs! You are right, that matchtigs themselves are not suitable for large graphs and dense graphs, as they require a lot of memory. Use the greedy heuristic instead, it is much more memory-friendly.
From: Enrico Rossignolo @.>
Sent: 22 June 2023 20:34
To: algbio/matchtigs @.>
Cc: Subscribed @.***>
Subject: Re: [algbio/matchtigs] std:bad_alloc
: How much memory do I need to compute matchtigs? (Issue #5)
Ok, the exception was due to blossom5 and not matchtigs. Maybe a warning about the memory requirement can be helpful since it requires O(|V|^2) if I'm correct.
— Reply to this email directly, view it on GitHubhttps://github.com/algbio/matchtigs/issues/5#issuecomment-1603060227, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AORQGR6BM262YWYQ4VK7JNLXMR6ZFANCNFSM6AAAAAAZQBPWPU. You are receiving this because you are subscribed to this thread.Message ID: @.***>
Thanks for the reply! I was just wondering how much better is the optimum solution wrt the one greedily computed 👍🏽
You can check the paper for some sample experiments. It is hard to estimate the amount of memory just from the number of nodes and edges. I have at least updated the documentation to warn about the high memory requirements for optimal matchtigs.
In general, also a large graph may be feasible to compute, it all depends on how many paths of length <= k (or possibly k-1, check the paper if needed ;) ) between nodes there are.
Hi! I'm having trouble processing a bcalm file with 12,585,517 nodes and 14,600,062 arcs (sequence SRR061958_1, 21-mer minimum count 1). I don't have enough RAM even if I've allocated 330GB for this job. How can I estimate how much memory I need?
The command and output are below