Direct proportional representation is easy enough. Just look at the number of votes each party got, and assign that percentage to their overall parliamentary representation. That roughly gives you the answer.

IRV is more interesting, but more complicated. It relies on some assumptions (e.g. Green, SNP, LibDem, Labour all preference each other 100%, Conservatives & Reform preference each other 100%) and takes a lot of effort to do on a seat-by-seat basis. And of course it all assumes ceteris paribus, when in actuality people would vote differently if the voting system were different.

As one example, here's the seat of Tatton:

Under IRV, with the above assumptions, Labour's Ryan Jude would have won with 26,005 votes to Conservative Esther McVey's 25,904. But tweak those assumptions just slightly (give 90% of LibDem votes to Labour, 10% to Conservatives) and it could go the other way (26,365 CON to 25,544 LAB). There are dozens or scores of seats where these sorts of interesting hypotheticals can be asked and analysed. IRV is actually, in my opinion, the next-worst voting system after FPTP (if you exclude weird and rarely-used ones like approval voting, range voting, etc.), but it's one of the most interesting to do analysis with.

STV is an utterly impossible comparison to make, because it relies on multi-seat electorates, which would probably be done by merging existing electorates into groups of 3–8. STV is a more generalised case of IRV so if you decided on how to do those merges, then you can get even more interesting analysis. As one example, if we imagine a merged electorate involving West Ham, East Ham, Ilford North, Ilford South, Leyton & Wanstead, and Stratford & Bow. Some assumptions are necessary to make this work, my assumption is that anyone whose party name says "workers" or "socialist" preferences Green and then Labour, while those mentioning religion preference Conservative, and if I don't know, I'm giving them to LibDems then Labour. I'm also assuming all voters for a named party vote as a block, preferencing the same candidate 1st, 2nd, etc., while independents get the votes as they were actually given. This is somewhat realistic because ballot paper design can be set up to encourage this in an STV context (see how Australia does it with "above the line" voting in the Senate, for an example). I've merged the minor parties named "workers" or "socialist" into a single party.

This is Labour heartland and in FPTP Labour won all 6 of these seats. My calculation ends up with 3 Labour, 1 LibDem, 1 Conservative, and 1 Green.

MMP ends up with basically the same overall result as direct proportional, but can be interesting in terms of independents & very minor parties and resulting overhang seats.

Given, many systems require more than just marking one box. While, even those that do not, would drastically change how people choose their vote.

I am unsure any such site can give a realistic result from available data?

Edit:

If we just assume proportional based on % of vote yesterday.

Tor 22.9% Lab 35.2% LDe 11.3% Ref 14.5% SNP 2.5% Oth 13.6%

It is bloody hard to see how either party could form a viable 50% Lab LD SNP and a few independents would take it over 50. But honestly, it is hard to imagine that working with the politicians voted yesterday. Tory Ref would need all independents. So less likely to work.

But as I said. Voters would go to the polls with very different ideas about how to vote.

There's this comment