Welcome to basic introduction to item values with respect to gold per sec items.
Here I will try to calculate the relative monetary values of 3 items (specifically Phil Stone, HoG, and Kage's Pick), and define to you the cost effectiveness of these items.
I also made a guide on Soraka within competitive play, ASSUMING you will always be at bot lane with a teammate (AD carry) to babysit.

As you read along, you will no doubt have questions or critiques; PLEASE READ THE EXTRA NOTES ON THE BOTTOM BEFORE CRITICIZING THE CALCULATIONS AND VALUES.

Greater Mark of Scaling Ability Power for added heals later on in the game
OR
Greater Mark of Armor for added defense against AD in bot lane because, in almost all competitive games, you will go against a support and an AD carry that can harass you and your teammate.

Greater Seal of Scaling Mana Regeneration for added mana regen throughout the game, because Soraka at level 18 can still run out of mana when you spam your E to silence enemies while spamming Q and W everytime it is up. At level 1-5 though, you shouldn't be running out of mana if your partner isn't being unnecessarily aggressive and you are positioning yourself right. Plus, you have health pots for yourself and 1 mana pot to persevere through early game.
OR
Greater Seal of Vitality for extra health as you level through the game. In combination with HoG, you will have sufficient health to survive small-time ganks.

Greater Glyph of Scaling Ability Power for added heals later on.

Greater Quintessence of Gold for your much needed gold gain from early to late game, especially since you will be giving the creep kills to your carry.
1 Greater Quintessence of Health for extra health at levels 1-5, just in case.

Pretty obvious at the choices. Only thing would be the defensive tree where I don't get dodge.
This is because I believe the health regen, although small, is more beneficial than the 2% increase in dodge, which, theoretically speaking, will be effective once every 50 hits from AD champions.

Other items to consider are Banshee's Veil, Aegis of the Legion, Rabadon's Deathcap and Randuin's Omen.

In general, I will try to avoid Banshee's Veil because it's most useful to champions that have channeling or long-cast ultis that don't want to be interrupted. It's also expensive, mainly because you are not rushing the Catalyst the Protector, and therefore will not benefit from its passive ability and will only be used as a substrate to complete the item. It is also why I get Mercury's Treads to somewhat compensate for the lack of Banshee's Veil.

Although Aegis of the Legion is a very good support item, I would usually leave it for the tank in the group to build it, especially since they will always be in the heart of the battle and give the aura to everyone on the team.

Rabadon's Deathcap is for when your team is already winning and you like to troll the opposing team in the face with a nuke-like heal, or when it's very late game and all your core items have finished.

I would like to point out the Randuin's Omen vs. Frozen Heart benefits. Although Randuin's Omen's active ability is more beneficial in counteracting AD carries (even with just 100 armor) up to around 5 seconds, you have to cast it at the right time and it can be misused. Some would also argue that for Frozen Heart to be effective, you have to be in the center of the fight in order to provide the debuff to everyone on the opposing team. I would argue, however, that the AD carries in the opposing team will not be focusing on our tanky champions who will be in the center of the battle, but rather target the squishy casters, of whom you will be beside most of the time. Therefore, the aura will almost always reach the carries and the passive will remain useful throughout the game.
Both items' stats are very beneficial, perhaps except for Randuin's health regen. While Randuin's Omen provides more defense in terms of health and the exclusive passive, Frozen Heart provides better casting with 500 mana and 20% CD reduction. Getting Frozen Heart also frees you from buying Ionion Boots of Lucidity and allows you get Mercury's Treads.

With the item build listed above, you will have 40% cd reduction without the added help of runes. If, however, you are thinking of building Randuin's over Frozen, then I suggest either Lucidity boots or cooldown glyphs.

In the end, it really boils down to the specific composition of your team vs their team, of your capability, and of your general playstyle.

Don't forget to always carry 3-5 wards everytime you leave the fountain so you can ward the jungle/river when your team is pushing, or when your team is getting pushed.

Again, pretty self-explanatory.
Only thing to note is the one point of Starcall which, I sometimes get at level 10 because I might have to defend a highly pushed lane with ONLY creeps when no other teammates are around.
Also I can spam it in teamfights while waiting for my CDs since the 8 magic reduction that stacks up to 20 times and lasts 8 sec is pretty useful.

If not, just ignore it completely till the end and max out E and W in a 1:1 fashion.

Before I drone on and on about the items, let's just start with the calculations right away and find out the "values" of these items with respect to their components. One thing to understand before we start is that these calculations are done exclusively with the item's components, and therefore is not an absolute account of how much the stats are worth.


kage's Pick
Kage's Lucky Pick

Components:
Amplifying Tome: 435g
Recipe: 330g
435 + 330 = 765g

Calculation of equivalency:
Since Kage’s Lucky Pick has 25 AP, it gained 25% more AP from Amplifying Tome. Therefore, you can think of the recipe cost as partly paying for the 25% increase, which means you can deduct this paid value from the total recipe cost to find out the “value” of the stat, 5gold/10sec.

25% of 435g is 108.75g.
330g – 108.75g = 221.25g (the value of 5gold/10sec stat)

Now simply calculate the time it would take to pay off 221.25g which is:
221.25g / 5g * 10sec /60sec = 7.375min to pay off


Heart of Gold


Components:
Ruby Crystal: 475g
Recipe: 350g
475g + 350g = 825g

Calculation of Equivalency:
Heart of Gold gives +70 health on top of the Ruby Crystal, which means it gives 70/180=0.3888 or 39% increase in health.
39% of 475g is 185.25g.
350g – 185.25g = 164.75g (the value of 5gold/10sec stat)

164.75g / 5g * 10sec / 60sec = 5.49min to pay off


Philosopher’s Stone
Philosopher's Stone

Components:
Faerie Charm: 180g
Regrowth Pendant: 435g
Recipe: 185g
180g + 435g + 185g = 800g

Calculation of Equivalency:
Philosopher’s Stone is very peculiar with respect to the other gold/sec items, and you will see why.

First, calculate the bonus mana regen gained from recipe.
+5 mana regen, which is 1.6666 times more than the Faerie Charm, or 167%.
167% of Faerie Charm’s cost is 180g * 1.67 = 300.6g
As you can see, the value of the amount of stat gained from creating the item already exceeds the recipe cost, making it effectively worth it even without the health regen and g/sec stats.
But, let’s continue on with the rest of the calculations.

+3 health regen is 20% of Regrowth Pendant, and so the value of the stat is worth 20% of 435g = 87g.
Now, deduct 300.6g + 87g from the recipe (185g), and you get -202.6g.

If we then proceed to calculate time of equivalency, we get:
-202.6g / 5g * 10sec / 60sec = -6.75min to pay off
Or inversely, you gained 6.75min worth of 5gold/10sec = 202.6g

*All the calculations are done under the assumption that you NEED these extra stats on the items, and not just the gold/sec stat. If you don’t need the extra stats, then technically the time of equivalency will be longer than stated because now you have to pay off for the gold you spent to get the useless stat.
Ex: If you don’t need the +5 AP gain from Kage’s Pick, or if you think “ehh, I don’t really need it but it’s not bad to be there”, then the recipe cost will not be deducted at all (or not be deducted as much if you thought the latter) and be somewhere between 330g and 220g.

*Calculations do not account for these items to be built into core items. This means, for example, if you are using Kage's pick to build into Deathfire Grasp, the value of Kage's pick would be higher (ie. instead of thinking 7.666min to pay off through 5g/10s, think of it as say 3-4min to pay off because you are not using this item purely for the intention of having 25AP and getting gold/sec )

*Theoretically speaking, the value of the gold/sec items also needs to take into account the possibility of having multiple routes to build higher tier items, which would effectively make it worth more since it gives you, the player, the option to have more adaptability, which is arguably worth some gold cost.

*Items such as Doran's blade, ring, and shield all provide extremely high amount of stats for the price, thereby being the single most cost effective items in the game.

*These calculations do not take into account the value of bag space (which is arguably worth more and more as the game goes on), and the fact that items with many components usually have cheaper recipes with lots of added stats because it occupies many precious bag space. If bag space was not the issue, having hundreds of Doran’s rings/swords/shields would be more cost effective over many higher tier items.

*Bag space can also be the reason why Philosopher's Stone is so cost effective, because it has 2 components instead of 1. However, even if this is the case, the high value of the Philosopher's Stone remains the same because in early game, the value of a bag space is next to nothing, especially if you are building Philosopher's Stone first.

*I believe item values also take into account how hard it is to build it. For example, Infinity Edge is an extremely cost effective item not because it has multiple components, but because it is hard (especially in competitive play) to farm up 1650g, which also stops you from getting other items that can be advantageous in the short term.
And thus it can be said that these items are long-term "investment" items.
Similar items are the "high-risk, high gain" items such as Bloodthirster and Mejai's Soulstealer.

*It's hard to calculate the value of some of more exclusive stats because some of them cannot be calculated in relative terms (such as Tiamat's splash effect; what is the monetary value in that?)

After considering ALL these non-accounted factors in calculation of the items’ worth, take what you will from this post, but personally, I believe the order of cost effectiveness should be like this:
Philo > HoG > Kage’s

I would recommend against Kage’s Lucky Pick for Soraka simply because having +25 AP is decent, but not highly needed, and so the payoff time would be even longer than 7 minutes. And on top of that, to get Deathfire Grasp the game must be very late game, around 50+ minutes, but even then I’d rather build rabadon’s deathcap because your healing will be highly boosted, and there is no magic resist for your heals whereas deathfire grasp’s dmg takes into account your magic penetration/reduction and the enemies magic resistance.

*If you are a tank that doesn’t require mana regen, then HoG should be the priority, because the majority of the negative cost of the philosopher’s stone is from the massive mana regen boost. Same goes for if you need health a lot more than health/mana regen, then definitely go HoG first.*

I understand the majority of this post can be intuitive, but I wanted to show one possible way of calculating the “worth” of items, especially in items where cost is weighed against the return.

It is critical to understand that it is near impossible to calculate the absolute value of items, because their value changes depending on the champions, situations, and time played in the game. The value also changes depending on the player due to playstyle and philosophy (such as the worth of bag space and exclusive stats) even with all factors being equal.