#pragma GCC optimize("O3")
#pragma GCC optimize("unroll-loops")
#include<iostream>
#include<string>
#include<cstdio>
#include<vector>
#include<cmath>
#include<algorithm>
#include<functional>
#include<iomanip>
#include<queue>
#include<ciso646>
#include<random>
#include<map>
#include<set>
#include<bitset>
#include<stack>
#include<unordered_map>
#include<unordered_set>
#include<utility>
#include<cassert>
#include<complex>
#include<numeric>
#include<array>
#include<chrono>
using namespace std;

//#define int long long
typedef long long ll;

typedef unsigned long long ul;
typedef unsigned int ui;
//ll mod = 1;
constexpr ll mod = 998244353;
//constexpr ll mod = 1000000009;
const int mod17 = 1000000007;
const ll INF = (ll)mod17 * mod17;
typedef pair<int, int>P;

#define rep(i,n) for(int i=0;i<n;i++)
#define per(i,n) for(int i=n-1;i>=0;i--)
#define Rep(i,sta,n) for(int i=sta;i<n;i++)
#define rep1(i,n) for(int i=1;i<=n;i++)
#define per1(i,n) for(int i=n;i>=1;i--)
#define Rep1(i,sta,n) for(int i=sta;i<=n;i++)
#define all(v) (v).begin(),(v).end()
typedef pair<ll, ll> LP;

using ld = double;
typedef pair<ld, ld> LDP;
const ld eps = 1e-10;
const ld pi = acosl(-1.0);

template<typename T>
void chmin(T& a, T b) {
	a = min(a, b);
}
template<typename T>
void chmax(T& a, T b) {
	a = max(a, b);
}
template<typename T>
vector<T> vmerge(vector<T>& a, vector<T>& b) {
	vector<T> res;
	int ida = 0, idb = 0;
	while (ida < a.size() || idb < b.size()) {
		if (idb == b.size()) {
			res.push_back(a[ida]); ida++;
		}
		else if (ida == a.size()) {
			res.push_back(b[idb]); idb++;
		}
		else {
			if (a[ida] < b[idb]) {
				res.push_back(a[ida]); ida++;
			}
			else {
				res.push_back(b[idb]); idb++;
			}
		}
	}
	return res;
}
template<typename T>
void cinarray(vector<T>& v) {
	rep(i, v.size())cin >> v[i];
}
template<typename T>
void coutarray(vector<T>& v) {
	rep(i, v.size()) {
		if (i > 0)cout << " "; cout << v[i];
	}
	cout << "\n";
}
ll mod_pow(ll x, ll n, ll m = mod) {
	if (n < 0) {
		ll res = mod_pow(x, -n, m);
		return mod_pow(res, m - 2, m);
	}
	if (abs(x) >= m)x %= m;
	if (x < 0)x += m;
	//if (x == 0)return 0;
	ll res = 1;
	while (n) {
		if (n & 1)res = res * x % m;
		x = x * x % m; n >>= 1;
	}
	return res;
}
//mod should be <2^31
struct modint {
	int n;
	modint() :n(0) { ; }
	modint(ll m) {
		if (m < 0 || mod <= m) {
			m %= mod; if (m < 0)m += mod;
		}
		n = m;
	}
	operator int() { return n; }
};
bool operator==(modint a, modint b) { return a.n == b.n; }
bool operator<(modint a, modint b) { return a.n < b.n; }
modint operator+=(modint& a, modint b) { a.n += b.n; if (a.n >= mod)a.n -= (int)mod; return a; }
modint operator-=(modint& a, modint b) { a.n -= b.n; if (a.n < 0)a.n += (int)mod; return a; }
modint operator*=(modint& a, modint b) { a.n = ((ll)a.n * b.n) % mod; return a; }
modint operator+(modint a, modint b) { return a += b; }
modint operator-(modint a, modint b) { return a -= b; }
modint operator*(modint a, modint b) { return a *= b; }
modint operator^(modint a, ll n) {
	if (n == 0)return modint(1);
	modint res = (a * a) ^ (n / 2);
	if (n % 2)res = res * a;
	return res;
}

ll inv(ll a, ll p) {
	return (a == 1 ? 1 : (1 - p * inv(p % a, a)) / a + p);
}
modint operator/(modint a, modint b) { return a * modint(inv(b, mod)); }
modint operator/=(modint& a, modint b) { a = a / b; return a; }
const int max_n = 1 << 20;
modint fact[max_n], factinv[max_n];
void init_f() {
	fact[0] = modint(1);
	for (int i = 0; i < max_n - 1; i++) {
		fact[i + 1] = fact[i] * modint(i + 1);
	}
	factinv[max_n - 1] = modint(1) / fact[max_n - 1];
	for (int i = max_n - 2; i >= 0; i--) {
		factinv[i] = factinv[i + 1] * modint(i + 1);
	}
}
modint comb(int a, int b) {
	if (a < 0 || b < 0 || a < b)return 0;
	return fact[a] * factinv[b] * factinv[a - b];
}
modint combP(int a, int b) {
	if (a < 0 || b < 0 || a < b)return 0;
	return fact[a] * factinv[a - b];
}

ll gcd(ll a, ll b) {
	a = abs(a); b = abs(b);
	if (a < b)swap(a, b);
	while (b) {
		ll r = a % b; a = b; b = r;
	}
	return a;
}
template<typename T>
void addv(vector<T>& v, int loc, T val) {
	if (loc >= v.size())v.resize(loc + 1, 0);
	v[loc] += val;
}
/*const int mn = 2000005;
bool isp[mn];
vector<int> ps;
void init() {
	fill(isp + 2, isp + mn, true);
	for (int i = 2; i < mn; i++) {
		if (!isp[i])continue;
		ps.push_back(i);
		for (int j = 2 * i; j < mn; j += i) {
			isp[j] = false;
		}
	}
}*/

//[,val)
template<typename T>
auto prev_itr(set<T>& st, T val) {
	auto res = st.lower_bound(val);
	if (res == st.begin())return st.end();
	res--; return res;
}

//[val,)
template<typename T>
auto next_itr(set<T>& st, T val) {
	auto res = st.lower_bound(val);
	return res;
}
using mP = pair<modint, modint>;
mP operator+(mP a, mP b) {
	return { a.first + b.first,a.second + b.second };
}
mP operator+=(mP& a, mP b) {
	a = a + b; return a;
}
mP operator-(mP a, mP b) {
	return { a.first - b.first,a.second - b.second };
}
mP operator-=(mP& a, mP b) {
	a = a - b; return a;
}
LP operator+(LP a, LP b) {
	return { a.first + b.first,a.second + b.second };
}
LP operator+=(LP& a, LP b) {
	a = a + b; return a;
}
LP operator-(LP a, LP b) {
	return { a.first - b.first,a.second - b.second };
}
LP operator-=(LP& a, LP b) {
	a = a - b; return a;
}

mt19937 mt(time(0));

const string drul = "DRUL";
string senw = "SENW";
//DRUL,or SENW
//int dx[4] = { 1,0,-1,0 };
//int dy[4] = { 0,1,0,-1 };

//------------------------------------

template<typename T>
struct BIT {
private:
	vector<T> node; int n;
public:
	void init(int n_) {
		n = n_; node.resize(n, 0);
	}
	//0-indexed
	void add(int a, T w) {
		for (int i = a; i < n; i |= i + 1)node[i] += w;
	}
	//[0,a)
	T sum(int a) {
		T ret = 0;
		for (int i = a - 1; i >= 0; i = (i & (i + 1)) - 1)ret += node[i];
		return ret;
	}
	//[a,b)
	T sum(int a, int b) {
		return sum(b) - sum(a);
	}
};

struct edge {
	int to;
};
using edges = vector<edge>;
using Graph = vector<edges>;
struct HLDecomposition {
	struct Chain {
		int depth;
		P parent;//chain number,index
		vector<P> child;//child chain number,parent index
		vector<int> mapfrom;
		BIT <modint> bt;

		Chain() { ; }
		Chain(int n) {
			bt.init(n);
		}
	};
	Graph baseG;
	vector<Chain> chains;
	vector<P> mapto;//raw index->chain number &index
	vector<vector<int>> mapfrom;//chain number & index ->raw index

	HLDecomposition() { ; }
	HLDecomposition(const Graph& g) {
		baseG = g;
		const int n = baseG.size();
		mapto = vector<P>(n, P{ -1,-1 });
		mapfrom.clear();
		vector<int> sz(n, 0);
		int start = 0;
		//int start = -1;
		//rep(i, n)if (baseG[i].size() <= 1) { start = i; break; }
		//assert(start != -1);
		size_check_bfs(start, sz);
		decomposition(start, start, 0, 0, 0, sz);
	}
	int depth(int t) {
		return chains[mapto[t].first].depth;
	}

private:
	void size_check_bfs(int start, vector<int>& sz) {
		const int n = baseG.size();
		queue<P> que;
		que.push({ start,start });
		int cnt = 0; vector<int> ord(n, -1);
		while (!que.empty()) {
			int from, parent;
			tie(from, parent) = que.front(); que.pop();
			ord[cnt++] = from;
			for (edge e : baseG[from]) {
				if (e.to == parent)continue;
				que.push({ e.to,from });
			}
		}
		//assert(cnt == n);
		reverse(all(ord));
		rep(i, n) {
			int from = ord[i];
			sz[from] = 1; for (edge e : baseG[from])sz[from] += sz[e.to];
		}
	}
	int decomposition(int from, int parent, int depth, int pnumber, int pindex, const vector<int>& sz) {
		vector<int> seq;
		bfs(from, parent, seq, sz);
		const int c = chains.size();
		chains.push_back(Chain((int)seq.size()));
		//chains.push_back(Chain());
		chains[c].depth = depth;
		chains[c].parent = { pnumber,pindex };
		rep(i, seq.size()) {
			mapto[seq[i]] = { c,i };
			chains[c].mapfrom.push_back(seq[i]);
		}
		mapfrom.push_back(chains[c].mapfrom);
		rep(i, seq.size()) {
			for (edge e : baseG[seq[i]]) {
				if (mapto[e.to].first != -1)continue;
				int nc = decomposition(e.to, seq[i], depth + 1, c, i, sz);
				chains[c].child.push_back({ nc,i });
			}
		}
		return c;
	}
	void bfs(int from, int parent, vector<int>& seq, const vector<int>& sz) {
		for (;;) {
			seq.push_back(from);
			int best = -1, next = -1;
			for (edge e : baseG[from]) {
				if (e.to == parent)continue;
				if (best < sz[e.to]) {
					best = sz[e.to]; next = e.to;
				}
			}
			if (next == -1)break;
			parent = from; from = next;
		}
	}
	vector<pair<int, P>> all_edge(int u, int v) {
		vector<pair<int, P>> res;
		if (depth(u) > depth(v))swap(u, v);
		while (depth(v) > depth(u)) {
			res.push_back({ mapto[v].first,{ 0,mapto[v].second + 1 } });
			P par = chains[mapto[v].first].parent;
			v = mapfrom[par.first][par.second];
		}
		while (mapto[v].first != mapto[u].first) {
			res.push_back({ mapto[v].first,{ 0,mapto[v].second + 1 } });
			P par = chains[mapto[v].first].parent;
			v = mapfrom[par.first][par.second];
			res.push_back({ mapto[u].first,{ 0,mapto[u].second + 1 } });
			par = chains[mapto[u].first].parent;
			u = mapfrom[par.first][par.second];
		}
		P p = minmax(mapto[v].second, mapto[u].second);
		res.push_back({ mapto[v].first,{ p.first + 1,p.second + 1 } });
		return res;
	}
	vector<pair<int, P>> all_vertice(int u, int v) {
		vector<pair<int, P>> res;
		if (depth(u) > depth(v))swap(u, v);
		while (depth(v) > depth(u)) {
			res.push_back({ mapto[v].first,{ 0,mapto[v].second + 1 } });
			P par = chains[mapto[v].first].parent;
			v = mapfrom[par.first][par.second];
		}
		while (mapto[v].first != mapto[u].first) {
			res.push_back({ mapto[v].first,{ 0,mapto[v].second + 1 } });
			P par = chains[mapto[v].first].parent;
			v = mapfrom[par.first][par.second];
			res.push_back({ mapto[u].first,{ 0,mapto[u].second + 1 } });
			par = chains[mapto[u].first].parent;
			u = mapfrom[par.first][par.second];
		}
		P p = minmax(mapto[v].second, mapto[u].second);
		res.push_back({ mapto[v].first,{ p.first,p.second + 1 } });
		return res;
	}

public:
	void add(int u, int a) {
		int id = mapto[u].first;
		int loc = mapto[u].second;
		chains[id].bt.add(loc, a);
	}
	modint query(int u, int v) {
		vector<pair<int, P>> vs = all_vertice(u, v);
		modint res = 0;
		rep(i, vs.size()) {
			int id = vs[i].first;
			int l = vs[i].second.first; int r = vs[i].second.second;
			res += chains[id].bt.sum(l, r);
		}
		return res;
	}
};

struct lcagraph {
private:
	int n;
	vector<vector<int>> G;
	vector<vector<int>> parent;
	vector<int> depth;
	int root;
	int tmp;
public:
	lcagraph(int n_) {
		n = n_;
		G.resize(n);
		parent.resize(n);
		depth.resize(n);
		tmp = 0;
		int cop = n;
		while (cop) {
			tmp++; cop /= 2;
		}
		rep(i, n)parent[i].resize(tmp);
		root = 0;
	}
	lcagraph() {}
	void init(int n_) {
		n = n_;
		G.resize(n);
		parent.resize(n);
		depth.resize(n);
		tmp = 0;
		int cop = n;
		while (cop) {
			tmp++; cop /= 2;
		}
		rep(i, n)parent[i].resize(tmp);
		root = 0;
	}
	void add_edge(int a, int b) {
		G[a].push_back(b);
		G[b].push_back(a);
	}
	void dfs(int id, int fr, int d) {
		parent[id][0] = fr;
		depth[id] = d;
		rep(j, G[id].size()) {
			int to = G[id][j];
			if (to == fr)continue;
			dfs(to, id, d + 1);
		}
	}
	void complete(int r = 0) {
		root = r;
		dfs(root, -1, 0);
		rep(j, tmp - 1)rep(i, n) {
			if (parent[i][j] < 0)parent[i][j + 1] = -1;
			else parent[i][j + 1] = parent[parent[i][j]][j];
		}
	}
	int lca(int u, int v) {
		if (depth[u] > depth[v])swap(u, v);
		for (int k = 0; k < tmp; k++) {
			if ((depth[v] - depth[u]) >> k & 1) {
				v = parent[v][k];
			}
		}
		if (u == v)return u;
		for (int k = tmp - 1; k >= 0; k--) {
			if (parent[u][k] != parent[v][k]) {
				u = parent[u][k];
				v = parent[v][k];
			}
		}
		return parent[u][0];
	}
	int dep(int x) {
		return depth[x];
	}
	int dist(int x, int y) {
		int l = lca(x, y);
		return depth[x] + depth[y] - 2 * depth[l];
	}
	//from a to b
	int proc_d(int a, int b, int d) {
		int l = lca(a, b);
		int res;
		if (dep(a) - dep(l) >= d) {
			res = a;
			rep(i, tmp) {
				if (d & (1 << i))res = parent[res][i];
			}
		}
		else {
			d = dep(a) + dep(b) - 2 * dep(l) - d;
			res = b;
			rep(i, tmp) {
				if (d & (1 << i))res = parent[res][i];
			}
		}
		return res;
	}
};


void solve() {
	int n, m; cin >> n >> m;
	vector<vector<int>> G(n);
	rep(i, m) {
		int a, b; cin >> a >> b; a--; b--;
		G[a].push_back(b);
		G[b].push_back(a);
	}
	vector<vector<int>> g(n);
	vector<vector<edge>> gg(n);
	vector<int> par(n);
	rep(i, n)par[i] = i;
	function<int(int)> find_par = [&](int id) {
		if (par[id] == id)return id;
		return par[id] = find_par(par[id]);
	};
	lcagraph lc(n);
	rep(i, n) {
		for (int to : G[i])if (i > to) {
			int p = find_par(to);
			if (p != i) {
				par[p] = i;
				g[i].push_back(p);
				gg[i].push_back({ p });
				gg[p].push_back({ i });
				lc.add_edge(i, p);
			}
		}
	}
	HLDecomposition hld(gg);
	lc.complete(n-1);
	vector<int> trans(n);
	int tmp = 0;
	vector<int> ri(n);
	function<void(int)> initdfs = [&](int id) {
		trans[id] = tmp; tmp++;
		for (int to : g[id]) {
			initdfs(to);
		}
		ri[id] = tmp;
	};
	initdfs(n-1);
	auto comp = [&](int a,int b) {
		return trans[a] < trans[b];
	};
	modint ans = 0;
	rep(i, n) {
		modint val = 1;
		vector<int> tos;
		for (int to : G[i]) {
			if (i > to)tos.push_back(to);
		}
		sort(all(tos),comp);
		vector<int> rs;
		rs.push_back(i);
		for (int to : tos) {
			int l = lc.lca(rs.back(), to);
			while (rs.size()&&trans[l] <= trans[rs.back()] && trans[rs.back()] < ri[l]) {
				rs.pop_back();
			}
			rs.push_back(l);
			int dif = lc.dep(to) - lc.dep(l);
			/*if (dif == 0) {
				cout << "! " << i << "\n";
				coutarray(tos);
			}*/
			assert(dif > 0);
			//cout << "? " << dif << "\n";
			int p = lc.proc_d(to, l, dif - 1);
			val += hld.query(to, p);
			rs.push_back(to);
			//cout << "?? " << to << " " << p << " " << hld.query(to, p) << "\n";
		}
		ans += val;
		hld.add(i, val);
		//cout << val << "\n";
		//cout << "?\n";
		//coutarray(tos);
	}
	cout << ans << "\n";
}



signed main() {
	ios::sync_with_stdio(false);
	cin.tie(0);
	//cout << fixed<<setprecision(10);
	//init_f();
	//init();
	//while(true)
	//expr();
	//int t; cin >> t; rep(i, t)
	solve();
	return 0;
}