jtackm
commented
1 year ago Hi Huizhen,
Thanks for the report!
Regarding Q1: You are correct: setting max_k=3 results in the inferred network only including direct associations (checked against up to three confounding variables). But these direct associations can also involve environmental factors: for instance, if a species directly depends on low pH or a particular habitat (independent of other species), this is useful information that FlashWeave tries to infer. There are currently no helper functions for filtering only otu-otu edges directly, but the network object has a meta_mask field that specifies, which variables are OTUs and which are MVs, which then together with the internally stored graph object can be used to filter for what you want. These are the functions I have been using internally:
import Graphs: ne, nv
using SimpleWeightedGraphs, FlashWeave
otu_ids(fw_res::FlashWeave.FWResult) = names(fw_res)[otu_indices(fw_res)]
mv_ids(fw_res::FlashWeave.FWResult) = names(fw_res)[mv_indices(fw_res)]
otu_indices(fw_res::FlashWeave.FWResult) = findall(.!meta_variable_mask(fw_res))
mv_indices(fw_res::FlashWeave.FWResult) = findall(meta_variable_mask(fw_res))
function otu_otu_edges(fw_res::FlashWeave.FWResult)
oids = Set(otu_indices(fw_res))
Iterators.filter(x->x.src in oids && x.dst in oids, SimpleWeightedGraphs.edges(graph(fw_res)))
end
function otu_mv_edges(fw_res::FlashWeave.FWResult)
oids = Set(otu_indices(fw_res))
Iterators.filter(x->begin
src_is_oid = x.src in oids
dst_is_oid = x.dst in oids
xor(src_is_oid, dst_is_oid)
end,
SimpleWeightedGraphs.edges(graph(fw_res)))
end
function mv_mv_edges(fw_res::FlashWeave.FWResult)
mvids = Set(mv_indices(fw_res))
Iterators.filter(x->x.src in mvids && x.dst in mvids, SimpleWeightedGraphs.edges(graph(fw_res)))
end
ne_otu_otu(fw_res::FlashWeave.FWResult) = length([true for x in otu_otu_edges(fw_res)])
ne_otu_mv(fw_res::FlashWeave.FWResult) = length([true for x in otu_mv_edges(fw_res)])
ne_mv_mv(fw_res::FlashWeave.FWResult) = length([true for x in mv_mv_edges(fw_res)])
ne(fw_res::FlashWeave.FWResult) = SimpleWeightedGraphs.ne(graph(fw_res))
nv(fw_res::FlashWeave.FWResult) = SimpleWeightedGraphs.nv(graph(fw_res))Perhaps they are useful to you? I may eventually add them to the package (not sure if they are intuitive enough?), but it would be good to at least highlight the role of the meta_mask and internal graph object in the documentation.
Regarding Q2: the "detailed" flag should be provided to save_network() (see README.md), not learn_network(). But the error message could be more clear, sorry about that.
Hope this helps, Janko
huizhen-yan
Hi, I'm stuck with the following two issues. Q1: Which edges in the FlashWeave results are environmentally driven? It appears that there is no clear indication in the learn_network() help or results. We can get the directed network by setting max_k=3, or univariate network (including direct and indirect edges) by setting max_k=0. In my opinion, the edges that do not appear in the direct network should include both spurious co-occurrence between microbes and environmentally-driven associations. However, I could not find any information on how to distinguish between these two parts in corresponding article and method documentation. Is it possible for FlashWeave to extract environmentally-driven edges?
Q2: FlashWeave runs normally when 'detailed' is not set, but I get the following error when setting 'detailed=true'. `julia> netw_results = learn_network(data_path, meta_path, sensitive=true, heterogeneous=false, alpha=0.001,max_k=3,detailed=true)
Can you help me? Thanks a lot.
Loading data
Normalizing
Removing variables with 0 variance (or equivalently 1 level) and samples with 0 reads -> no samples or variables discarded
Normalization
Removing meta variables with 0 variance (or equivalently 1 level) -> no samples or variables discarded
Learning interactions
Inferring network with FlashWeave - sensitive (conditional)
Automatically setting 'n_obs_min' to 20 for enhanced reliability Computing univariate associations
Univariate degree stats: Summary Stats: Length: 42362 Missing Count: 0 Mean: 738.558425 Minimum: 3.000000 1st Quartile: 382.000000 Median: 586.000000 3rd Quartile: 855.000000 Maximum: 4754.000000
Starting conditioning search
Preparing workers..
Done. Starting inference..
Exception occurred on worker 1: MethodError: no method matching check_candidate!(::Int64, ::Int64, ::Matrix{Float32}, ::Vector{Int64}, ::OrderedCollections.OrderedDict{Int64, Tuple{Float64, Float64}}, ::FlashWeave.FzTestCond{Float32, FlashWeave.NoNz}, ::Int64, ::Float64, ::Int64, ::Int64, ::Int64, ::Int64, ::Dict{Int64, Tuple{Tuple{Int64, Vararg{Int64}}, FlashWeave.TestResult, Tuple{Int64, Float64}}}, ::Bool, ::Vector{Int32}, ::Char, ::Bool, ::OrderedCollections.OrderedDict{Int64, Tuple{Float64, Float64}}; detailed=true) Closest candidates are: check_candidate!(::Int64, ::Int64, ::AbstractMatrix{ElType}, ::Vector{Int64}, ::OrderedCollections.OrderedDict{Int64, Tuple{Float64, Float64}}, ::TestType, ::Integer, ::AbstractFloat, ::Integer, ::Integer, ::Integer, ::Integer, ::Dict{Int64, Tuple{Tuple{Int64, Vararg{Int64}}, FlashWeave.TestResult, Tuple{Int64, Float64}}}, ::Bool, ::Vector{DiscType}, ::Char, ::Bool, ::OrderedCollections.OrderedDict{Int64, Tuple{Float64, Float64}}; bnb, cut_test_branches) where {ElType<:Real, DiscType<:Integer, TestType<:FlashWeave.AbstractTest} at ~/.julia/packages/FlashWeave/UprmH/src/hiton.jl:80 got unsupported keyword argument "detailed" Stacktrace: [1] kwerr(::NamedTuple{(:detailed,), Tuple{Bool}}, ::Function, ::Int64, ::Int64, ::Matrix{Float32}, ::Vector{Int64}, ::OrderedCollections.OrderedDict{Int64, Tuple{Float64, Float64}}, ::FlashWeave.FzTestCond{Float32, FlashWeave.NoNz}, ::Int64, ::Float64, ::Int64, ::Int64, ::Int64, ::Int64, ::Dict{Int64, Tuple{Tuple{Int64, Vararg{Int64}}, FlashWeave.TestResult, Tuple{Int64, Float64}}}, ::Bool, ::Vector{Int32}, ::Char, ::Bool, ::OrderedCollections.OrderedDict{Int64, Tuple{Float64, Float64}}) @ Base ./error.jl:165 [2] hiton_backend(T::Int64, candidates::Vector{Int64}, data::Matrix{Float32}, test_obj::FlashWeave.FzTestCond{Float32, FlashWeave.NoNz}, max_k::Int64, alpha::Float64, hps::Int64, n_obs_min::Int64, max_tests::Int64, prev_accepted_dict::OrderedCollections.OrderedDict{Int64, Tuple{Float64, Float64}}, candidates_unchecked::Vector{Int64}, time_limit::Float64, start_time::Float64, debug::Int64, whitelist::Set{Int64}, blacklist::Set{Int64}, rej_dict::Dict{Int64, Tuple{Tuple{Int64, Vararg{Int64}}, FlashWeave.TestResult, Tuple{Int64, Float64}}}, track_rejections::Bool, z::Vector{Int32}, phase::Char; fast_elim::Bool, no_red_tests::Bool, support_dict::OrderedCollections.OrderedDict{Int64, Tuple{Float64, Float64}}, kwargs::Base.Pairs{Symbol, Bool, Tuple{Symbol}, NamedTuple{(:detailed,), Tuple{Bool}}}) @ FlashWeave ~/.julia/packages/FlashWeave/UprmH/src/hiton.jl:138 [3] interleaving_phase(::Int64, ::Vararg{Any}; add_initial_candidate::Bool, univar_nbrs::OrderedCollections.OrderedDict{Int64, Tuple{Float64, Float64}}, kwargs::Base.Pairs{Symbol, Bool, Tuple{Symbol}, NamedTuple{(:detailed,), Tuple{Bool}}}) @ FlashWeave ~/.julia/packages/FlashWeave/UprmH/src/hiton.jl:155 [4] si_HITON_PC(T::Int64, data::Matrix{Float32}, levels::Vector{Int32}, max_vals::Vector{Int32}, cor_mat::Matrix{Float32}; test_name::String, max_k::Int64, alpha::Float64, hps::Int64, n_obs_min::Int64, max_tests::Int64, fast_elim::Bool, no_red_tests::Bool, FDR::Bool, weight_type::String, whitelist::Set{Int64}, blacklist::Set{Int64}, univar_nbrs::OrderedCollections.OrderedDict{Int64, Tuple{Float64, Float64}}, prev_state::FlashWeave.HitonState{Int64}, debug::Int64, time_limit::Float64, track_rejections::Bool, cache_pcor::Bool, kwargs::Base.Pairs{Symbol, Bool, Tuple{Symbol}, NamedTuple{(:detailed,), Tuple{Bool}}}) @ FlashWeave ~/.julia/packages/FlashWeave/UprmH/src/hiton.jl:338 [5] interleaved_worker(data::Matrix{Float32}, levels::Vector{Int32}, max_vals::Vector{Int32}, cor_mat::Matrix{Float32}, edge_rule::String, nonsparse_cond::Bool, shared_job_q::Distributed.RemoteChannel{Channel{Tuple}}, shared_result_q::Distributed.RemoteChannel{Channel{Tuple}}, GLL_args::Dict{Symbol, Any}) @ FlashWeave ~/.julia/packages/FlashWeave/UprmH/src/interleaved.jl:30 [6] (::FlashWeave.var"#122#126"{String, Bool, Matrix{Float32}, Vector{Int32}, Vector{Int32}, Matrix{Float32}, Dict{Symbol, Any}, Distributed.RemoteChannel{Channel{Tuple}}, Distributed.RemoteChannel{Channel{Tuple}}})() @ FlashWeave /Applications/Julia-1.8.app/Contents/Resources/julia/share/julia/stdlib/v1.8/Distributed/src/macros.jl:83 [7] #invokelatest#2 @ ./essentials.jl:729 [inlined] [8] invokelatest @ ./essentials.jl:726 [inlined] [9] #153 @ /Applications/Julia-1.8.app/Contents/Resources/julia/share/julia/stdlib/v1.8/Distributed/src/remotecall.jl:425 [inlined] [10] run_work_thunk(thunk::Distributed.var"#153#154"{FlashWeave.var"#122#126"{String, Bool, Matrix{Float32}, Vector{Int32}, Vector{Int32}, Matrix{Float32}, Dict{Symbol, Any}, Distributed.RemoteChannel{Channel{Tuple}}, Distributed.RemoteChannel{Channel{Tuple}}}, Tuple{}, Base.Pairs{Symbol, Union{}, Tuple{}, NamedTuple{(), Tuple{}}}}, print_error::Bool) @ Distributed /Applications/Julia-1.8.app/Contents/Resources/julia/share/julia/stdlib/v1.8/Distributed/src/process_messages.jl:70 [11] run_work_thunk(rv::Distributed.RemoteValue, thunk::Function) @ Distributed /Applications/Julia-1.8.app/Contents/Resources/julia/share/julia/stdlib/v1.8/Distributed/src/process_messages.jl:79 [12] (::Distributed.var"#100#102"{Distributed.RemoteValue, Distributed.var"#153#154"{FlashWeave.var"#122#126"{String, Bool, Matrix{Float32}, Vector{Int32}, Vector{Int32}, Matrix{Float32}, Dict{Symbol, Any}, Distributed.RemoteChannel{Channel{Tuple}}, Distributed.RemoteChannel{Channel{Tuple}}}, Tuple{}, Base.Pairs{Symbol, Union{}, Tuple{}, NamedTuple{(), Tuple{}}}}})() @ Distributed ./task.jl:484
ERROR: "Interleaved error (see stacktrace above)" Stacktrace: [1] interleaved_backend(target_vars::Vector{Int64}, data::Matrix{Float32}, all_univar_nbrs::Dict{Int64, OrderedCollections.OrderedDict{Int64, Tuple{Float64, Float64}}}, levels::Vector{Int32}, max_vals::Vector{Int32}, cor_mat::Matrix{Float32}, GLL_args::Dict{Symbol, Any}; update_interval::Float64, variable_ids::Vector{String}, meta_variable_mask::Nothing, convergence_threshold::Float64, conv_check_start::Float64, conv_time_step::Float64, parallel::String, edge_rule::String, edge_merge_fun::typeof(FlashWeave.maxweight), nonsparse_cond::Bool, verbose::Bool, workers_local::Bool, feed_forward::Bool, kill_remote_workers::Bool) @ FlashWeave ~/.julia/packages/FlashWeave/UprmH/src/interleaved.jl:161 [2] infer_conditional_neighbors(target_vars::Vector{Int64}, data::Matrix{Float32}, all_univar_nbrs::Dict{Int64, OrderedCollections.OrderedDict{Int64, Tuple{Float64, Float64}}}, levels::Vector{Int32}, max_vals::Vector{Int32}, cor_mat::Matrix{Float32}, parallel::String, nonsparse_cond::Bool, recursive_pcor::Bool, cont_type::DataType, verbose::Bool, hiton_kwargs::Dict{Symbol, Any}, interleaved_kwargs::Dict{Symbol, Any}) @ FlashWeave ~/.julia/packages/FlashWeave/UprmH/src/learning.jl:148 [3] learn_graph_structure(target_vars::Vector{Int64}, data::Matrix{Float32}, all_univar_nbrs::Dict{Int64, OrderedCollections.OrderedDict{Int64, Tuple{Float64, Float64}}}, levels::Vector{Int32}, max_vals::Vector{Int32}, cor_mat::Matrix{Float32}, parallel::String, recursive_pcor::Bool, cont_type::DataType, time_limit::Float64, nonsparse_cond::Bool, verbose::Bool, track_rejections::Bool, hiton_kwargs::Dict{Symbol, Any}, interleaved_kwargs::Dict{Symbol, Any}) @ FlashWeave ~/.julia/packages/FlashWeave/UprmH/src/learning.jl:176 [4] LGL(data::Matrix{Float32}; test_name::String, max_k::Int64, alpha::Float64, hps::Int64, n_obs_min::Int64, max_tests::Int64, convergence_threshold::Float64, FDR::Bool, parallel::String, fast_elim::Bool, no_red_tests::Bool, weight_type::String, edge_rule::String, nonsparse_cond::Bool, verbose::Bool, update_interval::Float64, edge_merge_fun::typeof(FlashWeave.maxweight), tmp_folder::String, debug::Int64, time_limit::Float64, header::Vector{String}, meta_variable_mask::Nothing, dense_cor::Bool, recursive_pcor::Bool, cache_pcor::Bool, correct_reliable_only::Bool, feed_forward::Bool, track_rejections::Bool, all_univar_nbrs::Nothing, kill_remote_workers::Bool, workers_local::Bool, kwargs::Base.Pairs{Symbol, Bool, Tuple{Symbol}, NamedTuple{(:detailed,), Tuple{Bool}}}) @ FlashWeave ~/.julia/packages/FlashWeave/UprmH/src/learning.jl:255 [5] macro expansion @ ./timing.jl:463 [inlined] [6] learn_network(data::Matrix{Any}; sensitive::Bool, heterogeneous::Bool, max_k::Int64, alpha::Float64, conv::Float64, header::Vector{String}, meta_mask::BitVector, feed_forward::Bool, fast_elim::Bool, normalize::Bool, track_rejections::Bool, verbose::Bool, transposed::Bool, prec::Int64, make_sparse::Bool, make_onehot::Bool, max_tests::Int64, hps::Int64, FDR::Bool, n_obs_min::Int64, cache_pcor::Bool, time_limit::Float64, update_interval::Float64, parallel_mode::String, extra_data::Nothing, share_data::Bool, experimental_kwargs::Base.Pairs{Symbol, Bool, Tuple{Symbol}, NamedTuple{(:detailed,), Tuple{Bool}}}) @ FlashWeave ~/.julia/packages/FlashWeave/UprmH/src/learning.jl:559 [7] learn_network(data_path::String, meta_data_path::String; otu_data_key::String, otu_header_key::String, meta_data_key::String, meta_header_key::String, verbose::Bool, transposed::Bool, kwargs::Base.Pairs{Symbol, Real, NTuple{5, Symbol}, NamedTuple{(:sensitive, :heterogeneous, :alpha, :max_k, :detailed), Tuple{Bool, Bool, Float64, Int64, Bool}}}) @ FlashWeave ~/.julia/packages/FlashWeave/UprmH/src/learning.jl:347 [8] top-level scope @ REPL[39]:1
julia>`